<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Carmen Escobosaf on Medium]]></title>
        <description><![CDATA[Stories by Carmen Escobosaf on Medium]]></description>
        <link>https://medium.com/@carmen.escobosaf?source=rss-195173ca90f5------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/0*kPWILHd9LDja3Rgk</url>
            <title>Stories by Carmen Escobosaf on Medium</title>
            <link>https://medium.com/@carmen.escobosaf?source=rss-195173ca90f5------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 18:13:16 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@carmen.escobosaf/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[How I Built EGFR Precision Explorer replicating and adapting SMART Precision Cancer Medicine based…]]></title>
            <link>https://medium.com/@carmen.escobosaf/how-i-built-egfr-precision-explorer-replicating-and-adapting-smart-precision-cancer-medicine-based-02d2578eb9c1?source=rss-195173ca90f5------2</link>
            <guid isPermaLink="false">https://medium.com/p/02d2578eb9c1</guid>
            <category><![CDATA[coding]]></category>
            <category><![CDATA[lung-cancer]]></category>
            <category><![CDATA[website-development]]></category>
            <category><![CDATA[medicine]]></category>
            <category><![CDATA[genetic-mutation]]></category>
            <dc:creator><![CDATA[Carmen Escobosaf]]></dc:creator>
            <pubDate>Sun, 08 Nov 2026 16:16:27 GMT</pubDate>
            <atom:updated>2026-11-11T14:46:43.133Z</atom:updated>
            <content:encoded><![CDATA[<h3>How I Built EGFR Precision Explorer replicating and adapting SMART Precision Cancer Medicine based on EGFR Mutations</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*09IUfCNLF2uPt960.jpg" /><figcaption>Image retrieved from: <a href="https://drmathangi.com/egfr-gene-mutation-lung-cancer-treatment/">https://drmathangi.com/egfr-gene-mutation-lung-cancer-treatment/</a></figcaption></figure><p>Why do lung cancer treatments stop working, even though the treatment was designed to target it and at first it worked?</p><p>An explanation to why this happens is linked to the genetic information of the cancer cells. Some genes can have mutations that can affect how much a cell can grow and respond to treatment and a specific example of a gene that can be mutated would be the EGFR gene. This gene is frequent in non-small cell lung cancer, NSCLC.</p><p>The EGFR gene is important in controlling cell growth and how much it can divide, so if any mutations occured in this gene, this could cause cells to grow abnormally which can contribute to cancer. And these genetic mutations can also change the way the cancer cells respond to the targeted treatment.</p><p>I started to become more interested in how EGFR mutations and treatment resistence were linked, and I wanted to start learning in depth about how different mutations in the EGFR gene related to non small cell lung cancer and why some treatments can become ineffective.</p><p>This is what made me want to recreate the EGFR Precision Explorer, which is an interactive website that helps doctors and patients understand the different types of genetic mutations in that gene, as well as a genomic exploration.</p><h3><strong>Learning about the different types of genetic mutations of EGFR in NSCLC</strong></h3><p>Before starting the website, I needed to understand better the three main mutations that happen in the EGFR gene which I wanted to include in my website: L858R, Exon 19 Deletion and T790M.</p><p>Even though these three mutations happen in the same type of gene, each one has a different process and works in a different way. Also, it was important that I understood this good enough to be able to explain it in the simplest way possible so people like patients, doctors and people interested in the topic could understand.</p><h3>Planning the EGFR Precision Explorer</h3><p>After researching and learning more about these mutations, I started to plan how I would then show the information and how I would structure the website. <br>I thought it would be good that the website had different sections you could click on so it would be easier to navigate. After thinking how many sections there would be, I put them in a coherent order based on the complexity or how detailed each section was going to be (from the most general sections to the most specific ones with the most complex data).</p><p>I divided the website into 5 main sections:</p><ol><li>Overview: I explained what is EGFR, what the website is about and what NSCLC meant.</li><li>Mutation explorer: In that section the person can select to understand and learn about three different EGFR mutations.</li><li>Genomic data: There, you can upload a CVS file filled in with research dataset that may come from a study or a set of medical data. Once you upload the CVS file, it allows to see the information in charts, tables using Plotly and shows an explanation for what the data in the file mean.</li><li>Treatment Resistance: It explains how new genetic mutations in the gene can make cancer cells grow resistant to the once-working targeted treatment.</li><li>Research sources: It gives links from good websites, scientific databases and things to learn further about the topic.</li></ol><figure><img alt="" src="https://cdn-images-1.medium.com/max/298/1*mECcFVxywdSiaY2l4nf5ag.png" /></figure><p>I wanted the website to be interactive so the people could not only read information, which they can do it by searching on Internet, but adding the interactive parts like the Genomic Data section which they can learn about mutations from a dataset themselves, not needing assistance from an expert.</p><h3>Building the website with Streamlit and Python</h3><p>To create the website EGFR Precision Explorer, I used resources like Python and Streamlit.</p><p>I used Python to organise the information about the different genetic mutations and also to be able to process the information of genomic datasets in the website. And I used Streamlit to be able to turn the Python code to an actual website that lets you interact and without having to build it from scratch.</p><p>In Streamlit, I used some commands and components to create different things for the website, like, <strong>st.sidebar.radio() </strong>and <strong>st.selectbox()</strong>. The first command, <strong>st.sidebar.radio()</strong>, is used to allow users to browse the five sections. On the other hand, the second command, <strong>st.selectbox()</strong>, is used to give a range of options to select the mutation the person wants to learn about.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/352/1*kEUdDqJ6-TRKAQ_59ITAPw.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/342/1*atyMIfjHXEEd1gFrF-nfWA.png" /></figure><p>I also used the components <strong>st.metric(),</strong> to make a summary of the information, and <strong>st.file_uploader(), </strong>so people are able to upload the CSV files.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/376/1*uuplynv1uGkEr2z3l8A9WQ.png" /></figure><p>For the visual parts of the website I used a custom called <strong>CSS</strong>. This made it more organised and made it look visually better by adding dark blue headings that stood out with a light background and in some sections there were white summary cards that showed the key information.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/974/1*rKfZBjjcipF7GR0PIFH4mQ.png" /><figcaption>Overview section</figcaption></figure><p>When I was doing the code, one of the most important stuff that I had to focus on was the <strong>MUTATIONS </strong>dictionary. The prupose of this dictionary was so I could save the information from the 3 different mutations in one place and, it was important so nothing that was written in the website was incorrect. Also, this helped me since each mutation has different characteristics and relation with treatment resistance, so it was easier and helped me to have it more organised to have stored all the information in the same place.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/982/1*pbdmm7uvnvOg78I9pqdlPQ.png" /></figure><h3>The Mutation Explorer</h3><p>This section of the website was used so people could select from the three mutations the one that they wanted to learn more about. Once the person selects the mutation, the website shows all the information about it.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZwfPAzXaibxBnkB_Sa9hlQ.png" /><figcaption>Mutation Explorer section</figcaption></figure><p>It shows the gene, exon and the type of mutation that is further explained with how it is changed, why it is important and how it interacts with the treatment. It also shows the probability of being resistant to targeted treatment.</p><p>I thought of this section so people could understand the differences between the mutations and how depending on which mutation a body has, it acts one way or another. For example, the mutation Exon 19 deletion happens because of a missing section in the DNA strand, while, L858R and T790M happens because of subsitutions in the EGFR protein.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*YzI3hXx6cI5uDyntzdG9VA.png" /><figcaption>Mutation Explorer section</figcaption></figure><p>Another thing I focused on was to separate the information into smaller parts so people could focus on one part and also so it’s more easy to read, instead of writing everything in a big paragraph. Finally, I added a small note at the end of the section that explain that the descriptions are simplified so they are easier to understand and the tumours or cancer cells response to treatment can change. This was important because the website is for doctors, patients or people that are learning about the topic or needs help to understand that the treatments that are used and the mutations can change between each case.</p><h3>Genomic data</h3><p>This was if not the most important part, one of the most important parts of the website.</p><p>This is because I wanted to put in my website an interactive part so the people could learn from real medical and genomic data instead of only reading about mutations which is something that you can figure out by searching it on the Internet. So, to make an interactive part for my website I decided to create a section where you can upload a CSV file and the person can learn about study data retrieved or real medical sample information.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*64GQ2RVmbjQki_ZQzfskRg.png" /><figcaption>Genomic data section</figcaption></figure><p>To do this section of my website I need to use <strong>Pandas</strong>, which is a component that comes from Python, so it is able to read the data and be able to explain it and process it into tables and graphs. The Pandas needs two specific columns called <strong>sample_id</strong> and <strong>mutation</strong>, but it also accepts other columns like <strong>cell_line</strong>, <strong>treatment</strong> and <strong>treatment_response</strong>. Once the person uploads the file, the code needs to check if those columns are there and if they’re not it will give error.</p><p>If the columns are present, the section gives a summary of the number of mutations, samples, treatment response and which columns are more relevant than others.</p><p>Another command I used in Panda was <strong>value_counts()</strong> that is used so the code is able to count how many times a mutation appear in the research dataset. I also used <strong>st.bar_chart()</strong> from Streamlit to later present the data in charts and tables so it’s more visual for people to understand.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/492/1*lILxw2B96vJoYDEnIBr7ng.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JtqoMYRtkm2LwkgURuhPAg.png" /><figcaption>Genomic data section with CSV file uploaded</figcaption></figure><p>I used <strong>pd.crosstab()</strong> in the case where there was treatment response data in the CSV file, to be able to compare the mutations based on how they responded to a specific treatment.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/418/1*nMj7Pbp94ZaRyVtrwo69dQ.png" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Dm0CdMc_AH5YP6TEM3JEUg.png" /></figure><p>Finally, if there was a <strong>cell_line column</strong>, the website gives you an option to select a cell line to learn about it. A cell line is a group of identical cells that can come from a living thing like a plant, human or animal that can be grown and laboratories keep them alive in a laboratory dish.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JQoGt6A_AHJo5pHPGLkccg.png" /></figure><h3>Treatment Resistance</h3><p>This section is also important since one of the main question of the topic I’m learning about is why treatment can become ineffective when it was targeted to the specific cancer cells.</p><p>Targeted treatments are used to attack the weakness of the cancer cells but sometimes genetic mutations can occur in tumours and can eventually make them resistant to the treatment.</p><p>With the Treatment Resistance section I wanted to show and explain people the way it works and how it can change depending on the mutations the tumour has. In the section there is an interactive part where the person can choose between the three EGFR mutations and learn how the affinity with the treatment changes. I made sure that the explanations were easy to understand and I did a part where the website gave a summary of the explanation all together.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*UNtQgB_BF6k4xasf8Xlcew.png" /><figcaption>Treatment resistance section</figcaption></figure><p>I thought of this section as a part to make people aware why it is so important to learn and study the mutations a cancer cell can go through and how it makes cancer harder to treat, as well as why precision medicine’s target is to understand about the genetic characteristics of a tumour.</p><h3>Research sources</h3><p>At the last section of the website, I added a part with other websites and sources so people can learn more about the topic and maybe go even deeper, into a more technical insight.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*DC54-20r0sd0MbTIsdzOWg.png" /><figcaption>Research sources section</figcaption></figure><p>Some websites I put in the section was the NCBI Gene, ClinVar, cBioPortal, COSMIC and SMART Precision Cancer Medicine.</p><p>I thought the last source was especially important since my whole website is inspired on SMART Precision Cancer Medicine, only that I made it focused on the EGFR gene, instead of cancer in general.</p><h3>The limitations and challenges of this website</h3><p>A big challenge I needed to face was to be able to understand the topic really well to be able to explain it in the most simple way.</p><p>This is because although it doesn’t seem like it, it is hard to transform complex terminology to simpler ones, especially when the topic is involving the DNA, the genetic mutations and how proteins work or change overtime. I had to investigate about the three different mutations in depth so I could then analyse what were the most important details I needed to explain to people through the website.</p><p>Another big challenge was doing the genomic data section. This is due to the section needing to be able to analyse, process and organise the information from an external dataset. I needed to make the code to be able to find out if there were missing columns and to analyse whether it would be able to continue analysing the dataset without the complete information. It was also really hard to manage and handle all of the different commands and how to apply them into my code without it giving any syntax error. Also, I had to use Streamlit to change my code into a visual website and it got difficult at times to share the code from two different apps to make the website.</p><p>Some limitations of the website is that the dataset that is going to be uploaded into my website isn’t going ever to be big enough so the results can’t be used to generalise to all patients that have Non-Small Cell Lung Cancer (NSCLC). This is because each patient has a different situation and each one has a different DNA structure and genetic information, so even though they may have the same cancer, it doesn’t mean they are the same since lung cancer is a heterogenous disease.</p><h3>What I learned</h3><p>With this website I’ve learned how to combine two different things, programming and biology.</p><p>I learned about how EGFR mutations can change the way cell grows and how it can cause cancer, like, how they relate to NSCLC. I also learned more and understood much better how targeted treatments work and why tumours can become resistant. A great skill I acquired is to be able to explain and learn the difference between understanding the topic in a technical way and later be able to explain it to someone else. I learned how if someone is able to explain it in the most simple way to another person, that means that someone actually master the topic.</p><p>I also learned a lot about how to code better and understood better how the coding syntaxis works. I also learned how to use Python, Streamlit, Pandas and GitHub which were all the resources I had to use to be able to create a website that wasn’t only good and interactive but also with organised information and be able to analyse a CSV file filled in with a dataset and create graphs, tables and explanations based on that.</p><p>Here is a link that takes you to the website EGFR Precision Explorer: <a href="https://egfr-precision-explorer-dvajywvoemaq9bdahkt634.streamlit.app/">https://egfr-precision-explorer-dvajywvoemaq9bdahkt634.streamlit.app/</a></p><p>Here is some inspiration I used to make some parts of the website: <br><a href="https://vicc.org/research/smart-pcm">https://vicc.org/research/smart-pcm</a></p><p><a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC6080684/">https://pmc.ncbi.nlm.nih.gov/articles/PMC6080684/</a></p><p><a href="https://apps.smarthealthit.org/app/smart-precision-cancer-medicine">SMART Precision Cancer Medicine</a></p><p><a href="https://medium.com/@carmen.escobosaf/how-can-egfr-mutations-make-non-small-cell-lung-cancer-harder-to-treat-4d61c56fd6d8">https://medium.com/@carmen.escobosaf/how-can-egfr-mutations-make-non-small-cell-lung-cancer-harder-to-treat-4d61c56fd6d8</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=02d2578eb9c1" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[How Do EGFR Mutations Make Non-Small Cell Lung Cancer Harder to Treat, and Can AI Help Predict…]]></title>
            <link>https://medium.com/@carmen.escobosaf/how-can-egfr-mutations-make-non-small-cell-lung-cancer-harder-to-treat-4d61c56fd6d8?source=rss-195173ca90f5------2</link>
            <guid isPermaLink="false">https://medium.com/p/4d61c56fd6d8</guid>
            <category><![CDATA[oncology]]></category>
            <category><![CDATA[genetic-mutation]]></category>
            <category><![CDATA[biotechnology]]></category>
            <category><![CDATA[medicine]]></category>
            <category><![CDATA[lung-cancer]]></category>
            <dc:creator><![CDATA[Carmen Escobosaf]]></dc:creator>
            <pubDate>Thu, 08 Oct 2026 14:22:48 GMT</pubDate>
            <atom:updated>2026-10-25T16:13:24.604Z</atom:updated>
            <content:encoded><![CDATA[<h3>How Do EGFR Mutations Make Non-Small Cell Lung Cancer Harder to Treat, and Can AI Help Predict Resistance?</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/600/0*a8GS87JRqSxL7Vyz" /></figure><p>Imagine being diagnosed with lung cancer and finding out that your tumour has a specific genetic mutation. Your doctors choose a treatment designed to target it. At first, it works. The tumour shrinks. But months later, the cancer starts growing again.</p><p>So here comes the question: If the treatment was designed to target the cancer’s weakness, how did the cancer escape it?</p><p>To understand how this can happen in lung cancer, we first need to look at what makes one lung tumour different from another.</p><h3>1.0 First: What makes one lung cancer different from another?</h3><p>Lung cancer is not a single disease, it is a lot of different types that can change depending on their characteristics, genetic features, and behaviour. This is known as a heterogeneous disease. These differences can happen not only between patients, but also between different cells inside the same tumour. This means that two people with lung cancer may have very different forms of the disease and may respond differently to treatment.[1]</p><p>This is one of the things that makes lung cancer so difficult to treat: there is no single version of the disease. Even within one tumour, different cells can behave differently.</p><p>However, lung cancer tumours can first be distinguished by how their cells and tissues look under a microscope (histological morphology). After taking a tumour biopsy, doctors examine the sample to identify the type of lung cancer.</p><figure><img alt="A diagram that shows the different factors that contribute to LC heterogeneity. (LINK)" src="https://cdn-images-1.medium.com/max/520/0*h4DtG9GvqyA0YeG6" /><figcaption>Diagram that shows the different factors that contribute to lung cancer heterogeneity. [3]</figcaption></figure><h4>1.1 Types of lung cancer:</h4><p>Lung cancer is mainly divided into two types: small-cell lung cancer (SCLC) and non-small cell lung cancer (NSCLC).</p><ul><li><strong>Small-cell lung cancer (SCLC):</strong> SCLC makes up around 15% of lung cancer cases. It grows quickly and often spreads early to other parts of the body through metastasis. [4]</li><li><strong>Non-small cell lung cancer (NSCLC):</strong> NSCLC makes up around 85% of lung cancer cases. It includes different types of tumours, with adenocarcinoma being the most common. Smoking is its main risk factor, although other factors can contribute. [5]</li></ul><p>An example of how lung cancers can differ at the cellular, molecular, and genetic levels is <strong>Adenocarcinoma</strong>.</p><p>Lung adenocarcinoma is the most common subtype of NSCLC. It can occur in both smokers and non-smokers and is usually found in the outer parts of the lungs. Its different characteristics are another example of tumour heterogeneity. [7]</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*doDYS5lx-Llq3yPk" /><figcaption>Diagram on types of lung cancer [6]</figcaption></figure><p><strong>Do different lung cancers have different genetic mutations? </strong>Yes, these genetic mutations can affect how quickly a tumour grows, how it spreads, and how it responds to treatment. [8]</p><p>Different lung tumours in NSCLC can contain mutations in genes such as EGFR, ALK, KRAS and ROS1. Identifying these mutations through genetic testing helps doctors choose suitable targeted treatments. [8][9]</p><p>In this article, I will focus on EGFR and explain why mutations in this gene matter when treating lung cancer.</p><p>After all this, one important question that should be asked is: <strong>Why are genetic mutations important when treating lung cancer?</strong></p><p>Genetic mutations are important because different lung tumours can have different genetic changes. These changes can affect how the cancer grows and how it responds to treatment. Genetic testing helps doctors find these mutations and choose a targeted treatment that is designed to act on specific changes in the tumour. [10]</p><p>But to fully understand why these mutations matter, we first need to understand what a mutation actually changes inside a cell and what is the meaning behind genetic mutations in a lung cancer cell. What does a genetic mutation actually do to a lung cancer cell?</p><h3><strong>2.0 So, what does it actually mean for a lung cancer cell to have a genetic mutation?</strong></h3><h4><strong>2.1 Contribution from mutations to lung cancer</strong></h4><p>Mutations contribute to lung cancer by changing genes that control cell growth and division. Some mutations activate oncogenes, which can make cells grow and survive more than they should. Others switch off tumour suppressor genes, which normally help control abnormal growth. Over time, these changes can cause cells to grow uncontrollably and form a tumour. Different mutations can also contribute to tumour heterogeneity. [11]</p><p>Mutations can be inherited (germline mutations) or acquired during a person’s lifetime (somatic mutations). Inherited mutations are present from birth and can be passed from parents to children. Somatic mutations develop in individual cells and are not usually passed on to children. [12][13]</p><p>Genetic susceptibility doesn’t mean someone will definitely develop cancer, it means their genetic makeup can make certain cancers more likely.</p><p>But this article is about lung cancer, so, <strong>are the mutations found in lung cancer cells always present in the person’s other cells? </strong>[14]</p><p>No.</p><p>In lung cancer, many mutations found in tumour cells are somatic. This means they may be present in the tumour but not in other cells of the body. [14]</p><p>However, inherited mutations can be present in most cells from birth.</p><p>This brings us to EGFR, a gene that can be mutated in some lung cancers.</p><h3><strong>3.0 What is EGFR and what happens when it is mutated?</strong></h3><h4>3.1 EGFR</h4><p>EGFR (Epidermal Growth Factor Receptor) is a protein found on the surface of cells. It receives signals from growth factors and helps control cell growth, division and survival. [15]</p><p>When a growth factor binds to EGFR, it activates an enzyme called tyrosine kinase. This enzyme sends signals inside the cell by adding phosphate groups to proteins, a process known as phosphorylation. These signals activate pathways involved in cell growth and DNA synthesis.</p><p>EGFR normally has an important role in healthy cells. However, certain mutations can make it overactive, causing cells to receive too many growth signals.</p><p>What I find most interesting here is that EGFR is not normally harmful. It has an important role in healthy cells, but a mutation can turn the same growth signal into something that helps cancer grow.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/503/0*ORq7oSH1BudzwRJ4" /><figcaption>Diagram of process in cell’s membrane with EGFR [16]</figcaption></figure><h4><strong>3.2 How do activating EGFR mutations cause continuous growth signalling?</strong></h4><p>When an activating EGFR mutation is present, the receptor can become overactive and continue sending growth signals even when there is no normal signal telling it to grow. This keeps signalling pathways involved in cell growth, survival, DNA synthesis, and division active for too long. As a result, cells may continue growing and dividing when they shouldn’t, which can contribute to tumour development. [17]</p><h4><strong>3.3 Which EGFR mutations are most common in lung cancer?</strong></h4><p>The two most common EGFR mutations in lung cancer are [18]:</p><ul><li><strong>Exon 19 deletions:</strong> A section of DNA in exon 19 is missing. This changes EGFR activity and can cause continuous growth signalling.</li><li><strong>L858R mutation in exon 21:</strong> The amino acid leucine is replaced by arginine at position 858 of the EGFR protein. This change can make EGFR overactive, increasing signals for cell growth and division.</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/520/1*7I3wOjHvzORgcPNdSSnmrA.png" /><figcaption>Normal EGFR [62]</figcaption></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/511/1*A9_z6ORs9XkEP389VzkKnw.png" /><figcaption>Mutated EGFR [63]</figcaption></figure><h4><strong>3.4 Why are they important in NSCLC/adenocarcinoma?</strong></h4><p>These mutations are important in NSCLC, especially in lung adenocarcinoma, because they act as driver mutations that help cancer cells grow and survive. [19]</p><h4><strong>3.5 If this mutation is helping the cancer grow, why wouldn’t researchers simply block EGFR?</strong></h4><p>Understanding the role of EGFR led to a new approach to treatment: instead of attacking rapidly dividing cells more generally, researchers could target the altered pathway itself.</p><p>And, if researchers know what is driving the cancer, the next question seems obvious: can they target it?</p><p>This is where targeted therapy becomes important.</p><h3>4.0 <strong>Can we target the mutation?</strong></h3><h4>4.1 What is targeted therapy?</h4><p>Targeted therapy is a type of cancer treatment designed to act on specific molecules, usually proteins, that help cancer cells grow, divide or survive. Unlike treatments that affect many rapidly dividing cells, targeted therapies focus on particular molecular changes in cancer cells. [20]</p><p><strong>Now, how does it work?</strong></p><ol><li>Doctors test a tumour sample to identify mutations or biomarkers, such as EGFR.</li><li>They select a targeted drug that acts on the mutation or the protein it affects.</li><li>The drug blocks or reduces the signals that help cancer cells grow.</li><li>Doctors monitor the response to treatment.</li><li>If the cancer becomes resistant, further testing may help identify new treatment options.</li></ol><p>In EGFR-mutated NSCLC, EGFR inhibitors target the abnormal growth signals that the cancer depends on. [21][23]</p><p>These drugs bind to the tyrosine kinase part of EGFR, reducing its activity and interrupting the signals that support cell growth and survival.</p><p>Two important signalling pathways are the MAPK pathway and the PI3K–AKT–mTOR pathway. Both help control cell growth, division and survival. Blocking EGFR can reduce the activity of these pathways, slowing cell proliferation and sometimes causing cancer cells to die. [22]</p><p>This explains why EGFR inhibitors can initially work well against tumours with mutations such as exon 19 deletions and L858R. However, the cancer can eventually develop resistance, meaning that the treatment becomes less effective.</p><p>This raises another question: what exactly are EGFR inhibitors?</p><h4>4.2 EGFR Inhibitors: switching off the growth signal</h4><p>What if a cancer cell had its “grow” signal stuck on? This is one way to understand what can happen in some cases of non-small cell lung cancer (NSCLC) with an EGFR mutation.</p><p>This is where EGFR inhibitors come in. These are drugs that block EGFR and stop it from sending these signals. One important type is called tyrosine kinase inhibitors (TKIs), such as gefitinib, erlotinib and osimertinib.</p><p>So, how do they actually stop the signal? [23]</p><ol><li>First, EGFR becomes activated. Normally, this happens when a growth factor, a molecule that tells cells to grow, attaches to the receptor.</li><li>Next, EGFR activates its tyrosine kinase part. A tyrosine kinase is a protein that helps pass signals inside the cell by adding phosphate groups to proteins. This is called phosphorylation. EGFR can also add phosphate groups to itself, called autophosphorylation, which helps switch the signal on.</li><li>Then, the signal moves through the cell using different signalling pathways. Two important ones are the MAPK pathway and the PI3K–AKT–mTOR pathway. These pathways help control cell growth, division and survival.</li></ol><p>EGFR inhibitors interrupt this chain. TKIs bind to the tyrosine kinase part of EGFR, stopping or reducing its activity. This means EGFR cannot pass the growth signal as easily.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*KE27H4w3oW-PWEFI.png" /><figcaption>How EGFR activation sends growth signals diagram [24]</figcaption></figure><p>The result is simple: the cancer cell gets fewer signals telling it to grow and survive. Cell proliferation, meaning cell division, can decrease, and the cell may enter cell-cycle arrest, where it stops dividing. If survival signals are reduced enough, some cancer cells may also die. [22]</p><p>This is especially important in EGFR-mutated NSCLC. Mutations such as exon 19 deletions and exon 21 L858R can make EGFR an important driver of cancer growth. When an EGFR inhibitor blocks this receptor, it takes away a signal that the cancer cell relies on.</p><p>The whole process can be summed up simply:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/738/1*cW6PpkgIoRKypMAePxy4tw.png" /><figcaption>EGFR inhibitor mechanism of action diagram [25]</figcaption></figure><p>So, EGFR inhibitors work by switching down a growth signal that has become too active in certain NSCLC cells. Instead of allowing the cancer cell to keep receiving the message to “grow and survive,” the treatment interrupts that message.</p><p>But this raises an obvious, and probably the most important question in this article:</p><h3><strong>5.0 Why does the treatment stop working?</strong></h3><p>This brings us back to the problem from the beginning: if EGFR can be successfully targeted, why could the cancer return?</p><p>EGFR-targeted treatments can initially shrink tumours, but cancer cells can change over time. A tumour may contain different groups of cells, some of which are more sensitive to treatment than others. [27][31][35]</p><p><strong>Treatment resistance</strong> occurs when cancer that was initially controlled by an EGFR inhibitor starts growing again even though treatment was given. [27][32][36]</p><p>Resistance can develop in different ways. The EGFR protein may change, or cancer cells may activate other signalling pathways, such as MET or HER2, to continue growing. [28][32][34]</p><p>Resistant cells may already exist before treatment, or new genetic changes may develop over time. Treatment removes sensitive cells while resistant cells survive and multiply. As a result, resistant cells can become a larger part of the tumour. [30][35]</p><p>This shows that treatment may still be targeting EGFR correctly, but the cancer has found another way to survive.</p><h4>5.1 But where do these resistant cells come from?</h4><p>One possibility is that the tumour already contained a small group of cells with changes that made them less sensitive to the drug. Another is that new genetic changes emerge as the cancer evolves. [30][35] Either way, treatment creates a kind of selective pressure: cells that cannot survive the drug are removed, while cells that can survive have a chance to multiply. Over time, those surviving cells can become a much larger part of the tumour. [27][35]</p><p>In a way, the treatment itself creates the conditions for the resistant cells to become more important. The cells that survive are the ones that now have the advantage.</p><p>This is one of the reasons cancer treatment is so difficult. The drug is not necessarily failing because it stopped hitting its target. The cancer may have changed what the target looks like, or found another route to continue growing. [27][32][36]</p><h4>5.2 The T790M example</h4><p>One well-known example of treatment resistance is the <strong>EGFR T790M mutation</strong>. Researchers found this mutation in tumours that stopped responding to earlier EGFR inhibitors, such as gefitinib and erlotinib. [29][31]</p><p>So, what does T790M actually do? The T790M mutation changes the structure of the EGFR protein in a way that increases its affinity for ATP, a molecule needed for the receptor’s kinase activity. First-generation EGFR inhibitors, such as gefitinib and erlotinib, compete with ATP for binding to EGFR. When the receptor binds ATP more strongly, these drugs become less effective at blocking its activity. As a result, EGFR can continue sending growth signals that help cancer cells survive and multiply. T790M has been found in around half of cases of resistance to first-generation EGFR inhibitors. [29][31]</p><p>However, T790M is not the only reason why treatment can stop working. Cancer cells can also find other ways to keep growing by using different genes and signalling pathways, such as MET, HER2, KRAS, BRAF and the PI3K pathway. In some cases, tumours can even change into a different type of cell through a process called histological transformation. [28][32][34][36]</p><p>This is where tumour heterogeneity becomes important. A tumour can contain different groups of cells, called subclones, which have different genetic changes. While some cells depend on EGFR and respond well to treatment, others may survive by using different pathways. [30][31][35]</p><p>This also explains why one biopsy may not show all the mutations in a tumour. Different areas of the same tumour can have different genetic changes. [30][33]</p><p>In other words, the cancer hasn’t simply become “immune” to the drug. It has changed the molecular conditions under which the drug has to work.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*fqiCKnFxzizaE0nFIbtUmw.png" /><figcaption>Mechanism of acquired T790M mutation and action of 3rd generation EGFR TKI. [26]</figcaption></figure><p>And T790M is only one example. Researchers have found several other mechanisms that can cause resistance to EGFR-targeted treatments. [27][32][34][35]</p><p>Osimertinib is a third-generation EGFR inhibitor that can target EGFR mutations, including T790M. It forms a covalent bond with an amino acid called cysteine at position 797 (Cys797) in the EGFR protein. This bond helps the drug block the receptor’s activity. However, another mutation, called C797S, changes this amino acid and can prevent osimertinib from forming the same bond. This can lead to resistance to osimertinib. Therefore, even when a treatment can overcome one resistance mutation, the cancer may develop another change that makes the treatment less effective. [65]</p><p>Clinical trial results also show why the development of newer EGFR inhibitors matters. In the FLAURA trial, patients with previously untreated advanced NSCLC with EGFR mutations who received osimertinib had a median progression-free survival of 18.9 months, compared with 10.2 months for patients who received gefitinib or erlotinib. Progression-free survival is the time a patient lives without the cancer getting worse or death occurring. These results showed that osimertinib could delay disease progression more effectively than the first-generation treatments used in the trial. However, resistance can still develop. [64]</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*cqIo7mftpOp8V8LHAPtUag.png" /><figcaption>Table 1. Comparison of first-, second- and third-generation EGFR tyrosine kinase inhibitors.</figcaption></figure><h4>5.3 What if the cancer finds another route?</h4><p>If blocking EGFR is no longer enough, cancer cells can sometimes rely on other signalling pathways. Researchers have identified several mechanisms of resistance involving genes and proteins such as MET, HER2, KRAS, BRAF and components of the PI3K pathway. [28][32][34][36] Some tumours can even change their biological identity through a process called histological transformation. [32][34][36]</p><p>This raises another question: <strong>how can the same tumour contain cells that respond differently to exactly the same treatment?</strong></p><p>The answer is tumour heterogeneity.</p><p>A tumour is not necessarily one uniform mass of identical cells. Instead, it can contain different subpopulations, or subclones, with different genetic changes. One group of cells might depend a lot on mutant EGFR and therefore be very sensitive to an EGFR inhibitor. Another group might have another alteration that allows it to survive when EGFR is blocked. [30][31][35]</p><p>Imagine removing almost every weed from a garden, only to discover that a few had survived because they were resistant to the treatment. With less competition around them, those survivors can spread again.</p><p>Cancer can behave in a similar progressive way. The treatment changes which cells have the advantage. [27][35]</p><p>This also explains why looking at only one tumour sample can sometimes give an incomplete picture. Different areas of the same cancer, or different tumours within the same patient, may contain different genetic changes. A biopsy from one location might miss a resistant population of cells somewhere else. [30][33]</p><p>So now we have a new problem.</p><p>If cancer can change during treatment, and different parts of the cancer can contain different genetic changes, how can researchers actually find out what has happened?</p><p>That is where sequencing becomes important. Instead of simply asking whether the tumour is growing or shrinking, researchers can look much more closely at its genetic material and search for the changes that may be allowing some cancer cells to survive. Next-generation sequencing and liquid biopsy can help researchers detect these molecular changes and monitor resistance. [30][33][35]</p><p>And that changes the question from “Why isn’t the drug working anymore?” to something much more useful:</p><p>“What has the cancer changed, and can we find a way to target it?”</p><h3>6.0 H<strong>ow do researchers find these genetic changes?</strong></h3><p>Researchers can compare tumour DNA collected before and after treatment to identify genetic changes linked to acquired resistance, such as EGFR T790M. [42]</p><p>Liquid biopsies analyse tumour DNA circulating in the blood. They are less invasive than tissue biopsies and can help researchers monitor changes over time. However, some tumours release very little circulating tumour DNA (ctDNA), so certain mutations may go undetected. [38][43]</p><p>Although NGS helps researchers identify many genetic changes, not all mutations are important for treatment resistance.</p><p>The problem is that sequencing gives researchers an enormous amount of information. Finding mutations is becoming easier, but working out which ones actually matter is the harder part.</p><p>This makes sequencing very useful for studying NSCLC treatment resistance. Researchers can see which mutations are present, compare them before and after treatment, and look for genetic changes that appear when the cancer becomes resistant. However, there is a problem: NGS produces a huge amount of genetic information. Not every mutation found in a tumour is important, and some mutations have an unknown effect. Researchers therefore need to work out which changes are actually linked to tumour behaviour and treatment resistance.</p><p>Finding the mutations is one thing, working out which patterns actually matter is another. And that is where AI becomes important.</p><h3><strong>7.0 Why bring AI into this?</strong></h3><p>NGS can produce large amounts of genetic data, which makes it difficult for researchers to identify the most important patterns. Artificial intelligence (AI), especially machine learning, can help researchers analyse this data and find connections between genetic changes, patient information and treatment results. [44]</p><p>For example, researchers could train a machine-learning model using genetic data and treatment results from patients with EGFR-mutated NSCLC. The model could look for patterns that might help predict which tumours are more likely to become resistant to EGFR inhibitors and stop responding to treatment.</p><p>However, finding a pattern does not mean that a mutation is the reason why treatment resistance develops. Researchers need to test the model using data from other patients to check whether its predictions are reliable and whether it can work well with new data. [44][45]</p><p>One method that researchers could use is Random Forest. This machine-learning method uses many decision trees and combines their results to make predictions or place data into different groups. Random Forest has been studied in cancer research to find patterns linked to how patients respond to treatment and how their disease develops over time. [46]</p><h3>8.0 <strong>What can Random Forest actually do?</strong></h3><p>Random Forest is a machine-learning method that combines the results of many decision trees to make a prediction. Each tree looks at different features in the data, and the model combines their results to make a final prediction.</p><p>In cancer research, these features could include genetic mutations, gene activity and information about patients. By looking at these factors together, Random Forest may find patterns linked to treatment response or how long patients survive. These patterns may be difficult to spot by looking at the data manually.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*dX50XlrJJeNQB4uS.jpg" /><figcaption>Diagram of how Random Forest works [61]</figcaption></figure><p>This can be useful when looking at mutations such as EGFR in NSCLC. Researchers already know that an EGFR mutation can affect how a patient’s cancer responds to certain treatments. However, EGFR is not the only thing that matters. A tumour can have other genetic changes, and treatment response can also depend on other factors. This is where machine learning can help by looking at all of this information together and finding patterns that might be difficult to notice just by looking at the data.</p><p>For example, instead of simply asking “Does this patient have an EGFR mutation?”, the AI could look at genetic mutations, gene expression and clinical information at the same time. It could then find patterns between these features and outcomes such as treatment response or patient survival. I think this is where the connection between genetic testing and AI becomes much clearer: sequencing can give researchers the information, while machine learning can help them make sense of it.</p><p>Researchers have already tested this idea using real lung cancer data. Wang and colleagues used Random Forest to study gene data from patients with lung adenocarcinoma. They found genes linked to how long patients lived and used them to group patients by risk.[47] This study shows that Random Forest can find patterns in lung cancer data. However, predicting how long a patient may live is not the same as predicting if a tumour will stop responding to EGFR inhibitors. The study still shows that Random Forest could be useful for future research into why some treatments stop working.</p><p>Besse and colleagues studied 298 patients with metastatic NSCLC who received immune checkpoint inhibitor (ICI) treatment. They tested different machine-learning methods to predict whether treatment would control the disease. Random Forest performed best among the methods they tested. The model used factors such as patients’ general health, haemoglobin levels and neutrophil-to-lymphocyte ratio. [48]</p><p>The model achieved an AUC of 0.74 and an accuracy of about 68%. This suggests that it could make useful predictions, but it did not always get them right. The researchers also said that the model needed to be tested on other patients.</p><p>Lind and Anderson used Random Forest and genetic data from the GDSC project to predict how cancer cells would respond to different drugs. One model achieved an AUC of 0.94. However, the study used cancer cells grown in laboratories rather than patients, and it did not focus specifically on resistance to EGFR inhibitors in NSCLC. [49]</p><p>However, there is still a problem. Just because two things are linked does not mean one causes the other. Random Forest can find links between genetic changes and treatment results, but more research is needed to understand why. I think this is important to remember when using AI in medicine: predicting what might happen is not the same as knowing why it happens.</p><p>So, if Random Forest is looking at so many different features, <strong>how does it know which ones are the most important?</strong></p><p>Random Forest can show which features are most important for its predictions. These features could include EGFR mutations, other genetic changes, gene expression or clinical information. One way to measure a feature’s importance is to randomly mix up its values and check whether the model becomes less accurate. If the model performs worse, this suggests that the feature is important for its predictions. However, feature importance does not prove that a feature causes treatment resistance. It only shows how useful that feature was for the model when making predictions from the data.</p><h3>9. Can we actually trust an AI prediction?</h3><p>Finding a pattern does not prove what caused it. In EGFR-mutated NSCLC, Random Forest may find genetic changes that are linked to treatment resistance, but researchers need more evidence to make sure these links are real. [50]</p><p>A model’s accuracy is also important. For example, the model studied by Besse and colleagues had an accuracy of about 68% and an AUC of 0.74. This means the model could make useful predictions, but it still made mistakes. [51]</p><p>One major problem is <strong>overfitting</strong>. This happens when a model learns the training data too well but struggles to make accurate predictions using new data. Researchers can lower this risk by testing the model on a separate group of patients who were not used to train it. This is called <strong>external validation</strong>. [52][53]</p><p>Another problem is <strong>bias</strong>. If the data used to train a model does not include enough people from different patient groups, the model may work better for some people than for others. Differences between hospitals, groups of patients and the ways data are collected can also affect how well the model works. [54]</p><p>Before AI models can help doctors make treatment decisions, researchers need to test them on new patients and in different hospitals and healthcare settings. Just because a model works well in one study does not mean it is reliable enough to be used in hospitals. [55][56]</p><p>Therefore, AI should be used to help with research and medical decisions, rather than replace doctors’ judgement.</p><h3>10. So where could this actually lead?</h3><p>So, could this approach eventually change how treatment resistance is managed?</p><p>One possibility is using genetic sequencing to find changes in a tumour while a patient is still receiving treatment.</p><p>If researchers can find changes linked to resistance earlier, they may be able to understand why a treatment is no longer working. This could help researchers study how NSCLC changes over time. [57]</p><p>Machine learning could help researchers find patterns linked to treatment resistance by looking at genetic changes and patient information together. This could help predict which tumours are more likely to stop responding to treatment and help doctors better understand cancer. [54]</p><p>AI could also help find new mutations linked to resistance by comparing tumour samples taken before and after treatment. By analysing large amounts of genetic data, AI could find patterns that are difficult for researchers to spot. However, finding a pattern does not prove that a mutation causes resistance, so more research is needed. [58]</p><p>Another possibility is combining genomic data with clinical data. Genomic data can show what is happening in the tumour, while clinical data can show how the patient is responding to treatment. Looking at both types of data could give researchers a clearer picture of why some patients respond to treatment and others do not.</p><p>This is one area where machine learning could be useful because it can look at many features at the same time. [59]</p><p>If these models become reliable enough, they could help with personalised cancer treatment. Doctors could use information about a patient’s tumour, genetic changes and clinical data to help choose a suitable treatment. However, the model would first need to work well with different patients and in different hospitals, not just with the data it was trained on. [53]</p><p>There are also ethical and privacy issues to consider. Genetic data is very personal, so it needs to be stored safely and used carefully. Researchers also need to think about who can access the data and whether the data includes different groups of patients. As AI becomes more common in cancer research, protecting patient information will be an important part of using it safely. [60]</p><p>Before these models could be used regularly in hospitals, they would need to be tested and validated. Researchers would need to check that they work with new patients, different hospitals and different groups of people. They would also need to show that the models actually help doctors make better decisions. A model that works well in one research study is not automatically ready to be used with patients. [50]</p><p><strong>Remember the patient at the beginning? </strong>The treatment worked, the cancer came back, and something had changed. We now know that the answer may not be as simple as one mutation causing everything. Cancer can change, treatment can put pressure on the tumour, and resistant cells can gain an advantage. Genetic sequencing can help researchers see those changes, while machine learning could help them find patterns in the huge amount of data.</p><p>Maybe the biggest challenge is not finding a treatment for cancer. It is finding ways to keep up when the cancer changes.</p><h3>My Perspective</h3><p>So, EGFR mutations, as we have seen, can help some NSCLC cancers grow. Targeted therapy can block the abnormal signals caused by these mutations, and at first, the tumour can shrink.</p><p>But cancer can change over time. Tumour cells can be different from each other, and some cells may survive the treatment. These cells can then grow and become more common. New mutations, such as T790M, or changes in other pathways can also allow the cancer to keep growing even when EGFR is being targeted.</p><p>And the problem is not that there are no treatments. The problem is that cancer can change and find new ways to survive. This makes treatment resistance one of the biggest challenges when treating EGFR-mutated NSCLC.</p><p>I think genetic sequencing and AI, especially Random Forest, could be useful for understanding treatment resistance in EGFR-mutated NSCLC. However, I also think it is very important to monitor changes in tumour DNA through liquid biopsies. Liquid biopsies can help researchers detect genetic changes as the cancer develops, while AI could help analyse this information and find patterns that may be linked to resistance. In my opinion, these methods could work better together than on their own. However, liquid biopsies can miss mutations when a tumour releases very little circulating tumour DNA, and AI predictions may be unreliable if the data used to train the model are limited or biased.</p><p>I would feel more confident about using AI to help make treatment decisions if studies showed that a model could accurately predict resistance in different groups of patients and at different hospitals, and that doctors could use its predictions to improve patient outcomes. Until then, I think AI should support medical research and doctors’ decisions rather than replace them. Overall, the future of treating EGFR-mutated NSCLC will depend on tracking how the cancer changes, identifying resistance mechanisms and testing whether new tools really help patients.</p><h3><strong>About Me:</strong></h3><p>I´m Carmen Escobosa Fernández. I´m a 15-year-old Spanish student interested in cancer biology, genetics and the use of artificial intelligence in medical research. I like to play sports (been playing volleyball 3 year now and currently playing tennis). I once did a presentation to the European Centre for Algorithmic Transparency (ECAT) about the phsycology behind social media from a teenager´s point of view, and participated as event manager in CIS SFP Youth-TEDx. I also enjoy learning new languages (I speak 3 non-native languages: English, French and Chinese). Through this article, I research and explain how EGFR mutations can contribute to treatment resistance in non-small cell lung cancer and how Random Forest could help researchers study this challenge.</p><h3>References</h3><p>[1] <a href="https://karger.com/hhe/article-abstract/83/2/65/161254/Lung-Cancer-One-Disease-or-Many">https://karger.com/hhe/article-abstract/83/2/65/161254/Lung-Cancer-One-Disease-or-Many</a></p><p>[2] <a href="https://pubmed.ncbi.nlm.nih.gov/29635240/">https://pubmed.ncbi.nlm.nih.gov/29635240/</a></p><p>[3] <a href="https://karger.com/pat/article/92/4/239/925667/Heterogeneity-of-Lung-Cancer-The-Histopathological">https://karger.com/pat/article/92/4/239/925667/Heterogeneity-of-Lung-Cancer-The-Histopathological</a></p><p>[4] <a href="https://www.nature.com/articles/s41572-020-00235-0">https://www.nature.com/articles/s41572-020-00235-0</a></p><p>[5] <a href="https://www.nature.com/articles/nrdp20159">https://www.nature.com/articles/nrdp20159</a></p><p>[6] <a href="https://www.lung.org/lung-health-diseases/lung-disease-lookup/lung-cancer/basics/lung-cancer-types">https://www.lung.org/lung-health-diseases/lung-disease-lookup/lung-cancer/basics/lung-cancer-types</a></p><p>[7] <a href="https://www.sciencedirect.com/science/article/abs/pii/S2173510724001447">https://www.sciencedirect.com/science/article/abs/pii/S2173510724001447</a></p><p>[8] <a href="https://www.healthline.com/health/nsclc/guide-to-lung-cancer-mutations">https://www.healthline.com/health/nsclc/guide-to-lung-cancer-mutations</a></p><p>[9] <a href="https://www.webmd.com/lung-cancer/story/nsclc-gene-mutations">https://www.webmd.com/lung-cancer/story/nsclc-gene-mutations</a></p><p>[10] <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC4489668/">https://pmc.ncbi.nlm.nih.gov/articles/PMC4489668/</a></p><p>[11] <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC3804875/">https://pmc.ncbi.nlm.nih.gov/articles/PMC3804875/</a></p><p>[12] <a href="https://www.sciencedirect.com/science/chapter/bookseries/abs/pii/S0065230X08605345">https://www.sciencedirect.com/science/chapter/bookseries/abs/pii/S0065230X08605345</a></p><p>[13] <a href="https://frontlinegenomics.com/what-is-non-genetic-inheritance/">https://frontlinegenomics.com/what-is-non-genetic-inheritance/</a></p><p>[14] <a href="https://www.cancer.gov/news-events/cancer-currents-blog/2022/cancer-mosaic-mutations-embryo">https://www.cancer.gov/news-events/cancer-currents-blog/2022/cancer-mosaic-mutations-embryo</a></p><p>[15] <a href="https://www.sciencedirect.com/science/article/pii/S0923753419640615">https://www.sciencedirect.com/science/article/pii/S0923753419640615</a></p><p>[16] <a href="https://www.researchgate.net/figure/Action-mechanism-of-epidermal-growth-factor-receptor-tyrosine-kinase-inhibitors_fig2_233915546">https://www.researchgate.net/figure/Action-mechanism-of-epidermal-growth-factor-receptor-tyrosine-kinase-inhibitors_fig2_233915546</a></p><p>[17] <a href="https://elifesciences.org/articles/65824">https://elifesciences.org/articles/65824</a></p><p>[18] <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC3804875/">https://pmc.ncbi.nlm.nih.gov/articles/PMC3804875/</a></p><p>[19] <a href="https://www.sciencedirect.com/science/article/abs/pii/S0031302516330434">https://www.sciencedirect.com/science/article/abs/pii/S0031302516330434</a></p><p>[20] <a href="https://www.cancer.gov/about-cancer/treatment/types/targeted-therapies">https://www.cancer.gov/about-cancer/treatment/types/targeted-therapies</a></p><p>[21] <a href="https://ar.iiarjournals.org/content/39/7/3341.short">https://ar.iiarjournals.org/content/39/7/3341.short</a></p><p>[22] <a href="https://www.sciencedirect.com/science/article/abs/pii/S0169500203001351">https://www.sciencedirect.com/science/article/abs/pii/S0169500203001351</a></p><p>[23] <a href="https://www.mdpi.com/2072-6694/11/3/366">https://www.mdpi.com/2072-6694/11/3/366</a></p><p>[24] <a href="https://www.researchgate.net/figure/An-illustrative-diagram-showing-the-mechanism-of-action-of-EGFR-inhibitors-as-anticancer_fig1_397867852">https://www.researchgate.net/figure/An-illustrative-diagram-showing-the-mechanism-of-action-of-EGFR-inhibitors-as-anticancer_fig1_397867852</a></p><p>[25] <a href="https://scispace.com/agents/pi3k-akt-ras-mapk-jak-stat-signaling-ce4a0e33">https://scispace.com/agents/pi3k-akt-ras-mapk-jak-stat-signaling-ce4a0e33</a></p><p>[26] <a href="https://www.researchgate.net/figure/Mechanism-of-acquired-T790m-mutation-and-action-of-3rd-generation-EGFR-TKI_fig3_313360076">https://www.researchgate.net/figure/Mechanism-of-acquired-T790m-mutation-and-action-of-3rd-generation-EGFR-TKI_fig3_313360076</a></p><p>[27] <a href="https://pubmed.ncbi.nlm.nih.gov/41219394/">https://pubmed.ncbi.nlm.nih.gov/41219394/</a></p><p>[28] <a href="https://pubmed.ncbi.nlm.nih.gov/41935667/">https://pubmed.ncbi.nlm.nih.gov/41935667/</a></p><p>[29] <a href="https://pubmed.ncbi.nlm.nih.gov/18227510/">https://pubmed.ncbi.nlm.nih.gov/18227510/</a></p><p>[30] <a href="https://pubmed.ncbi.nlm.nih.gov/29110836/">https://pubmed.ncbi.nlm.nih.gov/29110836/</a></p><p>[31] <a href="https://pubmed.ncbi.nlm.nih.gov/28002980/">https://pubmed.ncbi.nlm.nih.gov/28002980/</a></p><p>[32] <a href="https://pubmed.ncbi.nlm.nih.gov/35412115/">https://pubmed.ncbi.nlm.nih.gov/35412115/</a></p><p>[33] <a href="https://pubmed.ncbi.nlm.nih.gov/31416192/">https://pubmed.ncbi.nlm.nih.gov/31416192/</a></p><p>[34] <a href="https://pubmed.ncbi.nlm.nih.gov/28701107/">https://pubmed.ncbi.nlm.nih.gov/28701107/</a></p><p>[35] <a href="https://pubmed.ncbi.nlm.nih.gov/30718357/">https://pubmed.ncbi.nlm.nih.gov/30718357/</a></p><p>[36] <a href="https://pubmed.ncbi.nlm.nih.gov/36245822/">https://pubmed.ncbi.nlm.nih.gov/36245822/</a></p><p>[37] <a href="https://www.cancer.gov/types/lung/hp/non-small-cell-lung-treatment-pdq?utm_source=chatgpt.com">https://www.cancer.gov/types/lung/hp/non-small-cell-lung-treatment-pdq?</a></p><p>[38] <a href="https://www.cancer.gov/news-events/cancer-currents-blog/2017/liquid-biopsy-detects-treats-cancer">https://www.cancer.gov/news-events/cancer-currents-blog/2017/liquid-biopsy-detects-treats-cancer</a></p><p>[39] <a href="https://www.genome.gov/about-genomics/fact-sheets/DNA-Sequencing-Fact-Sheet">https://www.genome.gov/about-genomics/fact-sheets/DNA-Sequencing-Fact-Sheet</a></p><p>[40] <a href="https://www.cancer.gov/types/lung/hp/non-small-cell-lung-treatment-pdq">https://www.cancer.gov/types/lung/hp/non-small-cell-lung-treatment-pdq</a></p><p>[41] <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC4098820/">https://pmc.ncbi.nlm.nih.gov/articles/PMC4098820/</a></p><p>[42] <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC7755440/">https://pmc.ncbi.nlm.nih.gov/articles/PMC7755440/</a></p><p>[43] <a href="https://www.cancer.gov/news-events/cancer-currents-blog/2017/liquid-biopsy-detects-treats-cancer">https://www.cancer.gov/news-events/cancer-currents-blog/2017/liquid-biopsy-detects-treats-cancer</a></p><p>[44] <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC10025752/?utm_source=chatgpt.com">https://pmc.ncbi.nlm.nih.gov/articles/PMC10025752/</a></p><p>[45] <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC11210434/">https://pmc.ncbi.nlm.nih.gov/articles/PMC11210434/</a></p><p>[46] <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC8822229/">https://pmc.ncbi.nlm.nih.gov/articles/PMC8822229/</a></p><p>[47] <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC6729062/">https://pmc.ncbi.nlm.nih.gov/articles/PMC6729062/</a></p><p>[48] <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC8699503/">https://pmc.ncbi.nlm.nih.gov/articles/PMC8699503/</a></p><p>[49] <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC6622537/">https://pmc.ncbi.nlm.nih.gov/articles/PMC6622537/</a></p><p>[50] <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC13392822/">https://pmc.ncbi.nlm.nih.gov/articles/PMC13392822/</a></p><p>[51] <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC8699503/">https://pmc.ncbi.nlm.nih.gov/articles/PMC8699503/</a></p><p>[52] <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC8991704/">https://pmc.ncbi.nlm.nih.gov/articles/PMC8991704/</a></p><p>[53] <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC11843972/">https://pmc.ncbi.nlm.nih.gov/articles/PMC11843972/</a></p><p>[54] <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC13476218/">https://pmc.ncbi.nlm.nih.gov/articles/PMC13476218/</a></p><p>[55] <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC9261114/">https://pmc.ncbi.nlm.nih.gov/articles/PMC9261114/</a></p><p>[56] <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC13392822/">https://pmc.ncbi.nlm.nih.gov/articles/PMC13392822/</a></p><p>[57] <a href="https://www.cancer.gov/about-cancer/treatment/types/cancer-treatment-resistance">https://www.cancer.gov/about-cancer/treatment/types/cancer-treatment-resistance</a></p><p>[58] <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC13036062/">https://pmc.ncbi.nlm.nih.gov/articles/PMC13036062/</a></p><p>[59] <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC13547827/">https://pmc.ncbi.nlm.nih.gov/articles/PMC13547827/</a></p><p>[60] <a href="https://www.who.int/publications/i/item/9789240029200">https://www.who.int/publications/i/item/9789240029200</a></p><p>[61] <a href="https://consuledge.com.au/blog/mastering-random-forests-a-beginners-guide-to-smarter-machine-learning/">https://consuledge.com.au/blog/mastering-random-forests-a-beginners-guide-to-smarter-machine-learning/</a></p><p>[62] <a href="https://es.slideshare.net/slideshow/receptor-ppt-by-kashikant-yadav/81760581#29">https://es.slideshare.net/slideshow/receptor-ppt-by-kashikant-yadav/81760581#29</a></p><p>[63] <a href="https://www.sciencemedia.com/l/Oncology_Primer_Overview/">https://www.sciencemedia.com/l/Oncology_Primer_Overview/</a></p><p>[64] <a href="https://doi.org/10.1056/NEJMoa1713137">https://doi.org/10.1056/NEJMoa1713137</a></p><p>[65] <a href="https://pmc.ncbi.nlm.nih.gov/articles/PMC6088362/">https://pmc.ncbi.nlm.nih.gov/articles/PMC6088362/</a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4d61c56fd6d8" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>