Real-world data and explainable artificial intelligence. Advancing precision medicine
Thesis event information
Date and time of the thesis defence
Topic of the dissertation
Real-world data and explainable artificial intelligence. Advancing precision medicine
Doctoral candidate
Master of Science Gunjan Chandra
Faculty and unit
University of Oulu Graduate School, Faculty of Information Technology and Electrical Engineering, Biomimetics and Intelligent Systems Group
Subject of study
Computer Science and Engineering
Opponent
Associate Professor Sami Äyrämö, University of Jyväskylä
Custos
Docent Pekka Siirtola, University of Oulu
Role of data and understandable artificial intelligence for enabling personalised medicine
Precision medicine is about giving the right treatment to the right person at the right time. Instead of using the same approach for everyone, it uses information about each patient to support more personalised care. This thesis explores how everyday health data and transparent artificial intelligence can be used together to make this kind of care more practical and trustworthy.
The research consists of four studies. The first study looks at how artificial health data can be created and shared safely. It introduces a way to check whether such data still reflects real patient data accurately, making it possible to share information without risking patient privacy.
The second study focuses on a rare blood cancer. It shows that even when only limited data is available, computer models can help predict how severe a serious treatment complication might become. This demonstrates that modern data methods can also support care for rare diseases.
The third study examines people with type 2 diabetes. It looks at different long-term blood sugar patterns and uses patient information such as health status, treatments, and social background to predict which pattern a person is likely to follow. Importantly, the models are designed so that it is clear which factors influence each prediction.
The fourth study compares several diabetes medications and estimates how well each one is likely to work for an individual patient. By combining results from clinical trials with real-world patient data, the study helps bring research findings closer to everyday medical practice.
Overall, the thesis shows that simple, transparent AI tools can turn health data into useful and fair decision support for doctors. This supports better care for people with cancer, chronic illnesses, and rare diseases.
The research consists of four studies. The first study looks at how artificial health data can be created and shared safely. It introduces a way to check whether such data still reflects real patient data accurately, making it possible to share information without risking patient privacy.
The second study focuses on a rare blood cancer. It shows that even when only limited data is available, computer models can help predict how severe a serious treatment complication might become. This demonstrates that modern data methods can also support care for rare diseases.
The third study examines people with type 2 diabetes. It looks at different long-term blood sugar patterns and uses patient information such as health status, treatments, and social background to predict which pattern a person is likely to follow. Importantly, the models are designed so that it is clear which factors influence each prediction.
The fourth study compares several diabetes medications and estimates how well each one is likely to work for an individual patient. By combining results from clinical trials with real-world patient data, the study helps bring research findings closer to everyday medical practice.
Overall, the thesis shows that simple, transparent AI tools can turn health data into useful and fair decision support for doctors. This supports better care for people with cancer, chronic illnesses, and rare diseases.
Created 25.12.2025 | Updated 30.12.2025