AI in Disease Diagnostics

Debi Ahitov
4 min readNov 28, 2020


Hi everyone, in my first blog post of the series of “AI in Molecular Biology”, I want to introduce you all to AI in disease diagnostics.

As we all know, advancements in technology, and more specifically in AI, are rapidly reaching each field. One of the fields that greatly benefit from these developments in the health sector. From bioengineering to bioinformatics, all fields of the health sector are developing with the aid of technology.

One advancement AI has made in this sector is in the concept of disease diagnostics. The most common AI method used for disease diagnostics is Machine Learning. ML provides computer systems with the capability to automatically learn and improve from experience without being explicitly programmed. It focuses on computer programs developing so that they can access data and use it in order to learn by themselves. Therefore, it’s highly beneficial for making data analysis. Today, ML is suitable for analyzing medical data and more specifically, diagnosing diseases or developing appropriate drugs. In this blog post, we will concentrate on disease diagnostics.

If we were to classify ML, it could be separated into branches as supervised learning, unsupervised learning, and reinforcement learning. When a supervised learning model is used, the computer algorithm learns from a labeled data set, which provides an answer key that the algorithm can use to evaluate its accuracy on training data. On the contrary, an unsupervised model presents unlabeled data, from which the algorithm tries to take out the meaning by extracting features and patterns on its own. Finally, reinforcement learning models train an algorithm with a reward system, providing feedback when an artificial intelligence agent performs the best action in a particular situation.

As long as the science, and more specifically the medical, field has arisen, the scientists have been collecting data from the experiments. And when the most recent experiment data are looked for, many conclusions may appear as well. However, picking out certain data may be costly and time-consuming when people try to do it by themselves. Therefore, machines making conclusions and predictions with the help of AI has accelerated the pace of scientific processes.

I would like to mention a supervised learning project I have developed with my team in InspiritAI. Last summer, the project I had chosen was “Medical Imaging to Detect Pneumonia”. For this purpose, our aim was to develop a computer program that would detect whether a chest X-ray scan needed to be diagnosed with pneumonia or not. In order to work on this project, we had a data set filled with pictures and labels, which were previously diagnosed by doctors whether they consisted of pneumonia signs or not. By using this method, our project was successful since it mostly gave true positive and true negative results. These types of projects can be used in emergency rooms in hospitals. In fact, the system was programmed and tested in Boston, showing “promising results”.

Other applications present in our day are: detecting lung cancer/strokes based on CT scans, assessing the risk of sudden cardiac death or other heart diseases based on electrocardiograms and cardiac MRI images, classifying skin lesions in skin images, and finding indicators of diabetic retinopathy in eye images. These listed applications are very practical today since there’s plenty of data available in such cases and the algorithms are becoming just as accurate as diagnostics by experts. The difference is: the algorithm can draw conclusions in a fraction of a second, and it can be reproduced inexpensively all over the world. Soon everyone, everywhere could have access to the same quality of top experts in radiology diagnostics, and for a low price. However, more and more systems are present and being developed as more data are collected.

Consequently, the applications of ML in disease diagnostics are just initiated: more ambitious systems will develop in the next few years by involving a combination of data sources such as CT, MRI, genomics and proteomics, patient data, and even handwritten files. But don’t worry! AI won’t be able to replace doctors; instead, it will be used to mark potential dangers and they will be needed for necessary interpretations.