Kaiku Health and Oulu University Hospital have jointly published a research investigating the feasibility of machine learning in predicting symptoms of patients undergoing Immune Checkpoint Inhibitor (ICI) treatments. The study is the first of its kind and the results were published at the ESMO Immuno-Oncology Congress 2019 in Geneva on December 12.
The results indicate that machine learning-based modeling of patient-reported symptom data on ICI treated cancer patients can predict the onset and continuity of symptoms related to ICI toxicities. The best performance considering all metrics was found for shortness of breath, joint pain, cough and fatigue. Importantly, all 14 symptom prediction models performed at a good level.
The results are encouraging and have several practical applications.
“The results are a step forward in better utilizing the existing patient-reported outcomes data. If we are able to accurately predict symptoms related to ICI toxicities, it can help us intervene earlier, provide proactive support for patients and enable more precise follow-up”, Dr. Sanna Iivanainen, the main investigator of the study at Oulu University Hospital, summarises.
The Kaiku Health digital platform was used to capture symptom data in a real-world setting from patients undergoing ICI therapies. In the research, the anonymized and aggregated PRO data was used to train and tune models for symptom continuity and onset prediction for 14 symptoms related to ICI toxicities.
For more information, contact:
Chief Product Officer, Kaiku Health
+358 45 276 3593
About Kaiku Health
Kaiku Health is a digital health intervention platform classified as a Medical Device in cancer care. Its algorithms screen symptoms, notify care team and provide personalised support for patients. Kaiku Health has modules for over 25 cancer types across different cancer care pathways and is currently in use in over 40 European cancer clinics and hospitals.