Artificial intelligence (AI) is revolutionising how scientists study cancer and Type 1 diabetes and discover ways to fight them. From finding new drug targets and predicting who might develop disease, to tailoring treatments for individual patients, AI is reshaping biomedical research, diagnostics and therapy planning, making them faster, smarter and more precise.
Every year, cancer and Type 1 diabetes take a growing toll on people worldwide. In 2022, around 20 million new cases of cancer were diagnosed, and nearly 9.7 million lives were lost to the disease. Experts estimate that by 2050, new cancer cases could rise to more than 35 million, driven by ageing populations and lifestyle risks such as tobacco use, obesity and alcohol.
At the same time, Type 1 diabetes (T1D), in which the body’s immune system mistakenly attacks the cells that produce insulin, is also on the rise. In 2025, it is estimated that about 9.5 million people globally are living with this type of diabetes, up from 8.4 million in 2021.
Both diseases are rooted in the immune system but in opposite ways. In cancer, the immune system is too weak or shuts down, allowing tumours to grow unchecked. In Type 1 diabetes, the immune system goes into overdrive and attacks healthy cells. Despite this, both conditions stem from immune system imbalances—just in different directions, as Nobel laureate Peter Medawar first observed in his 1960 Nobel lecture.
READ I GST reforms unfinished without input tax credit fix
AI in drug discovery
Artificial intelligence is rapidly transforming the way new drugs and therapeutic targets are discovered in cancer immunology.
Traditionally, finding tumour antigens or immune checkpoints that could serve as drug targets has been a time-consuming process, requiring years of laboratory work. AI now speeds this up by sifting through huge datasets—from genetic sequences to single-cell immune profiles—to detect patterns that reveal how cancer cells escape immune surveillance.
For instance, machine-learning models can identify novel tumour antigens that are invisible to the human eye but crucial for immune recognition, and can also highlight immune checkpoint molecules that suppress T-cell activity—T cells being immune cells that detect and destroy infected or cancerous cells. Novel tumour antigens help the immune system recognize and attack cancer cells, while immune checkpoint molecules can inhibit this response, making them both critical targets in cancer immunotherapy. Machine learning can help identify hidden antigens important for immune targeting and pinpoint checkpoint proteins that weaken T-cell responses.
These insights are already being used by pharmaceutical companies: AstraZeneca, for example, has partnered with Immunai to apply AI-driven immune modelling to find biomarkers and guide drug dosing strategies in oncology trials.
Moreover, AI platforms integrating single-cell RNA sequencing data that contain information on the genetic messages inside cells with immune cell analysis have shown promise in predicting which tumour–immune interactions are most critical for therapy design. By automating the discovery of such targets, AI shortens the drug development pipeline, improves accuracy in selecting immune pathways to modulate, and opens the door to highly personalised immunotherapies.
Biomarker discovery
One of the biggest challenges in cancer immunology is determining which patients will actually benefit from immunotherapies such as checkpoint inhibitors, a type of cancer treatment that helps the immune system fight cancer by blocking off signals from checkpoint proteins, and allowing T-cells to kill the tumour. Not all tumours will respond to this therapy, and without reliable biomarkers, many patients face ineffective treatments and wasted time. AI helps solve this by analysing complex datasets such as medical imaging, genomic sequencing and immune signatures to identify predictors of response.
For example, deep learning models can help predict how well a person might respond to treatment by analysing things like how many mutations the tumor has and how many immune cells are nearby. They also check for PD-L1, a kind of “invisibility cloak” that cancer cells use to hide from the immune system and a key target for checkpoint immunotherapy, which helps unmask them.
In clinical studies, AI-driven analysis of medical images has been shown to predict how well patients will respond to a treatment that lifts the “brakes” off the immune system, allowing it to attack cancer cells, and it does this more accurately from CT scans than traditional methods. By combining multiple data types, AI is enabling more precise patient selection, reducing trial and error in immunotherapy, and ultimately making treatments safer and more cost-effective.
Immunotherapy optimisation
AI not only predicts who will respond to immunotherapy but also refines how treatments are designed and delivered. Boosting the immune system against tumours must be balanced with avoiding autoimmune diseases as a side effect of an overactive immune system.
AI models combine genetic information of patients (genomics data), protein data also known as proteomics, as well as other information from patient records to mimic the interaction of tumours with the immune system and predict and design efficient treatment strategies.
A recent deep-learning approach, for example, classified “hot” tumours (immune-rich, responsive) versus “cold” tumours (immune-poor, resistant) across cancers. These tools help doctors tailor therapy combinations such as checkpoint inhibitors with targeted drugs based on the tumour microenvironment, improving success rates while reducing toxicity for safer, personalised care.
Analysing the tumour ecosystem
Tumours are not uniform; they are complex ecosystems of cancer cells, immune cells and stromal cells interacting dynamically. To understand this complexity, researchers use single-cell sequencing and multiple layers of biological data, which generate massive datasets.
AI is essential here because it can process millions of data points per patient to uncover hidden patterns. For instance, neural-network-based algorithms have been applied to single-cell RNA sequencing data to classify immune cell states and reveal how they infiltrate tumours in cancers such as lung, colorectal, breast and pancreatic cancer.
Personalised cancer vaccines
Perhaps one of the most exciting areas where AI is making a difference is in the development of personalised cancer vaccines. These vaccines aim to train the patient’s immune system to recognise tumour-specific mutations called neoantigens.
Identifying the right neoantigens is a massive challenge, but AI models can predict which mutations are most likely to be presented on tumour cells and trigger a strong immune response.
Early clinical trials using AI-guided personalised vaccines have shown encouraging results: for example, recent studies in liver and kidney cancer patients demonstrated strong immune activation and even long-term remission after vaccination. With AI accelerating the design pipeline, the dream of customised cancer vaccines tailored to each patient’s tumour genetics is moving closer to routine clinical reality.
AI and Type 1 diabetes
In the case of diabetes, AI is helping detect T1D far before symptoms appear by analysing genetic using data from the TEDDY cohort—a long-term study tracking children at risk for type 1 diabetes—used machine learning to analyze genetic risk, immune markers, and blood samples from early infancy to predict who would develop the disease by age six. This model achieved strong accuracy, highlighting that combining early-life metabolism, genetics and immunity offers a powerful way to predict risk even in very young children.
AI is also uncovering which immune cells contribute most to destruction of the insulin-producing cells in the pancreas known as β-cells. For example, advanced analysis methods have discovered new groups of special immune fighter cells (known scientifically as CD4⁺ T cells) that are found in higher numbers in people with type 1 diabetes—something traditional analysis using older methods had missed.
Continuous glucose monitoring and management
Managing T1D is made easier by “artificial pancreas” systems—hybrid closed-loop devices that automatically adjust insulin using real-time CGM data. AI-driven systems learn each patient’s insulin needs and glucose patterns, delivering better control with less effort. Clinical trials show these systems maintain glucose in the target range more consistently and reduce dangerous lows compared with standard pumps.
AI is fundamental to these systems: it adapts insulin delivery based on individual data, optimising dosing dynamically.
AI is becoming a powerful partner in immunology because it can make sense of the complex data our immune system generates. By doing this, it speeds up discoveries that once took years, helps doctors design treatments that fit each person’s unique biology, and even finds connections between very different diseases—such as cancer, where the immune system is too weak, and Type 1 diabetes, where it is overactive. In both cases, AI is turning overwhelming data into clear insights, bringing us closer to earlier diagnosis, smarter therapies and truly personalised medicine.
Jugal Das is a Ramalingaswami Fellow, and Varsha Dhanda is a CSIR-NET qualified PhD student at the Shiv Nadar Institution of Eminence, Delhi- NCR. Originally published under Creative Commons by 360info
.