Deep learning systems have widely presented promising results in cancer diagnostics [1] and for Kaiku Health, machine learning (ML) based approaches are not anything new. Previously, Oulu University Hospital (OYS) and Kaiku Health have shown that it is possible to predict both symptom and adverse event onset and continuation in patients receiving immune checkpoint inhibitor (ICI) therapies. In our previous study, we investigated whether machine learning could be used to predict the treatment response, or more specifically objective response rate (ORR) of ICI-treated patients, based on clinical and patient-reported data [2].
A new study conducted with OYS brings us yet another step closer to predicting treatment outcomes of cancer patients for immune checkpoint inhibitor therapies with machine learning. It investigates whether an ML-based prediction model for immune-related adverse events (irAEs) of cancer patients receiving ICIs could be created based on patient-reported outcome (ePRO) symptom data coupled with clinical data.
The dataset utilized in the study had two sources. The first consisted of 820 completed symptom questionnaires from 34 ICI-treated advanced cancer patients, 18 monitored symptoms collected using the Kaiku Health digital platform. The second included prospectively collected irAE data, the Common Terminology Criteria for Adverse Events (CTCAE) class, and the severity of 26 irAEs. The ML models were furthermore built using extreme gradient boosting algorithms. The first model was trained to detect the presence and the second the onset of irAEs.
Results of the study. Performance metrics for the prediction models for the presence and onset of irAES.
The results are promising. The four metrics used in the study included accuracy score, Area Under the Curve (AUC), F1-score and Matthew’s correlation coefficient (MCC). Based on all four, the model trained to predict the presence of irAEs had an excellent performance.
The study suggests that using ePRO data as an input, ML based prediction models can indeed predict the presence and onset of irAEs with a high accuracy. This indicates that ePRO follow-up with ML algorithms could facilitate the detection of irAEs in ICI-treated cancer patients. Eventually, early detection of irAEs could result in improved toxicity profile and quality of life for cancer patients.
To learn more, contact us! You can also find the whole article in BMC Medical Informatics and Decision Making.
References
- McKinney SM, Sieniek M, Godbole V, et al. International evaluation of an AI system for breast cancer screening. Nature. 2020;577(7788):89–94. https:// doi. org/ 10. 1038/ s41586- 019- 1799-6.
- https://kaikuhealth.com/kaiku-health-has-taken-its-first-steps-to-predict-treatment-response-of-cancer-patients-for-immune-checkpoint-inhibitor-therapies-with-machine-learning/
