MY DATA SCIENCE VOYAGE— FROM A NOOB TO CREATING AN AI BASED HEALTHCARE APPLICATION

and what you can learn from me to start your AI journey

Introduction

The term ‘Artificial Intelligence’ or ‘Data Science’ has always been a representation for a new change in the tech industry. Over the years, with improvements on various different aspects on technology growth, things have taken a turn with the ‘AI hype’ that has caused curiosity in making career choices, product development and increase in start-ups. From various videos on educational websites to social media ads screaming AI onto us and tech giant companies delivering products mainly focused on AI, this has become a global outrage where there is an easy access to learning and building new products that can make it or break it in the industry and giving people a chance to become the next successful data scientist.

Now that I have given a brief overview on this new tech trend that started not too long ago, here is my journey on how I started from the dude who graduated engineering, but not knowing a basic for loop structure, to a developer who created an AI powered healthcare application that can help save a lot of lives from skin cancer.

The beginning of a venture

It all started back in 2019 when I stood there in front of a company on a bright sunny afternoon after my final university exams with formals and a petrified face when the HR called me into the room for my technical interview. To give a little insight on my interview preparation, there was just one word to explain it, ‘disappointing’. Yes, even after attending few interviews in the past, I didn’t have a proper study plan and got rejected due to my performance in coding. I didn’t understand simple coding problems and certain logical representations. With the pressure to appease the interviewer, I stammered and choked on my own answers after some basic SQL and java questions. A few days after the interview, I got the call from the HR congratulating me for getting selected as a fresher. This was the best day of my life. I was excited to see what the next phase was going to be like.

One evening, as I was in my cabin, a senior employee came up to me and introduced himself as a big data developer. After some conversations, he gave me few tasks to work on. This was an opportunity to strengthen my coding skills and learn python. With more tasks coming, I was able to sharpen my skills and gain confidence in the work I was doing. My thoughts became clear and so did my vision to see the road ahead.

Weeks passed and I started moving towards analytics. This included creating graphical representations to see how the data has evolved, finding missing values and preparing data, understanding feature engineering. Seeing everything from an analytical perspective was easy, also interesting to make crucial decisions and to prepare diagrammatic understandings of the upcoming tasks. However, my mentor made sure that the path of learning ML (machine learning) and DL (deep learning) was a challenging trip and I paid attention to details throughout the learning phase. I was keen to know about my new learnings that was in store.

A rough path

I started with the complete data science pipeline. This pipeline is defined as the flow of process that handles the data and gets features and labels from it for training any ML or DL model. The flow consists of preparing the data, feature engineering, feature selection, model training and deployment. My mentor started training me on classification and regression algorithms and how to solve problems using these techniques.

But things took a sharp turn as the pandemic came into existence in the early 2020 and all offices were shut down. As I was on leave, there was plenty of time for me to continue my learning. So I decided to self-study all the way. After getting through all the classification models, I started with a data science problem, sentiment analysis.

Rough path

Honestly, I should say this, it changed me for the better, although there were many obstacles on the way. It was hard to debug the code when there were issues, but keeping a steady mind and patience was the key factor for getting the solution out of a problem. There were sleepless nights and ten plus hours of work. But all of these gave results that I needed because I never gave up on the work I put in.

In the end, I created an application that could predict the sentiment of a foreign language and provide the results in an application. But there was still an incomplete feeling in my heart. I wanted to do something innovate which can be of use to many people. Something was telling me to push forward and do something unique. That led me to learn deep learning and change the way I was.

An idea to innovation

Few months ago, there was a competition on Kaggle organized by the Society for Imaging Informatics in Medicine (SIIM) on skin cancer. This was a perfect opportunity to learn deep learning and computer vision although the competition was a bit intimidating and challenging for a beginner. I could have probably taken a simple problem statement for starters, but this was definitely a challenge to learn the technology and solve such a complex problem. So, I took my time for understanding to handle the concepts of deep learning, computer vision. The journey of learning to handle the data, training, validating the model and learning new concepts like pytorch, keras was definitely an experience I will never forget. Looking through articles, research papers, websites, blogs, coding competition sites, researching into simplistic methods of solving code issues, increasing accuracy, avoiding overfitting the model, learning about performance metrics, fixing things on the way, creating a flask application and adding styling sheets and html pages, integrating the model to the application, sleepless nights, my brain going haywire, dreaming in my sleep of fixing the code.

Every day was a struggle to get things right. My focus on getting towards the finishing point was visualized shorter but felt like a million miles away. Your entire physiology changes when you work so hard that you never see time fly like that. There are times you feel dejected and the desire to give up on your work. Trust me, I was there. But the hunger to bring that change will always spread the heat and push you harder towards your goal. That’s how I accomplished an entire health application.

About my product

After months of work, I finally created my own product that can make a difference in many lives. That’s how I came up with an AI based healthcare application DermaChecker, an algorithmic skin care specialist that classifies skin images sent by users based on its cancerous state.
Here is how it works. The user sends the image of the skin mole to our application. DermaChecker provides the result based on the image observation. ‘Normal’ meaning benign and ‘at risk’ meaning malignant.

Please do visit DermaChecker get your results now and do share this with your loved ones if you find the application efficient.

My advice to you

· Never rush into learning something just because it is a “hype” or “it can bring you a lot of money”. Do your research on understanding what this role means to you and how comfortable are you with it.

· Once you take one foot in, Do not think about it. There will never be a ‘right time’ to start something.

· Surf into the basics of high school stats and programming (like python) to jumpstart your data science career.

· Understanding data and visualizing it with graphs and story-telling it is one of the factors to becoming successful at your job. Focus on EDA (Exploratory data analysis) and data preprocessing techniques to sharpen your skills

· Attend online competitions on Kaggle and use your tools to enhance the fluency of handling complex data. I believe that is a huge step in learning to handle data.

· Go through the machine learning models with a mathematical mindset. When I say ‘mathematical’, I mean understanding the basics of what the algorithm is about from a numeral standpoint. It gives you so much value to take decisions in using it during a problem statement.

· Both ML and DL can be a little intimidating at first, but always try and move slow. The knowledge will definitely embrace you.

· Finally, start making a live product to show your work and what you can do. Trust me, it’s a transformation to reckon with.

Conclusion

Dedication and perseverance are two important words to know from this. It is that continuous work that has bought me to where I am. From the guy who couldn’t understand loops or any coding technique to building a healthcare application is seriously an accomplishment I will hold on to and take pride in. I am just a beginner who has a lot to learn and prove to the world. Start your journey today and I wish you the best of luck!

Data Scientist, machine learning enthusiast, Software engineer, innovator