Artificial Intelligence for Prostate Cancer Diagnosis and Prognostication
Abstract
Our group, led by DDLS Fellow, Assistant Professor Kimmo Kartasalo, focuses on medical AI, particularly in application to cancer. The increasing adoption of digital pathology enables using AI for improved cancer diagnostics and ultimately better treatment decisions. We intend to develop diagnostic and prognostic AI systems for prostate cancer clinical management by integrating: large-scale datasets of digitized prostate biopsies and MRI scans from an international network of health care providers, follow-up information on clinical outcomes, and genomic profiling of the samples. In terms of fundamental methodology, we also advance model and instrument calibration approaches, prediction uncertainty estimation, and computational efficiency to support robust, safe and scalable medical AI.
These systems will allow improving the quality and efficiency of pathology reporting and ultimately extending prognostication beyond human capacity by directly predicting disease progress and optimal treatment from tissue images integrated with radiology and molecular data. We envision these approaches being capable of modeling the current state of individual patients and tumors based on their earlier clinical trajectory, serving as a basis for predictive reasoning communicated to clinicians via LLM-based interfaces.
Utilizing diverse international data and calibration methods will allow training robust models and evaluating them in view of real-world sources of variation. Moreover, estimating the reliability of predictions is key for handling outliers and artefacts which are unavoidable in real-world use of diagnostic AI, and for providing useful feedback to the medical experts using the software in the clinic. We plan on validating the AI solutions in real-world environments by conducting prospective clinical trials. This will initially focus on diagnosis and grading of prostate cancer, and direct prognostication of outcomes and LLM reasoning in the second phase.
