AI Model MSI-SEER Advances Cancer Prediction with Uncertainty Quantification

Breakthrough in Cancer Prediction: MSI-SEER AI Model

Researchers from the USA and Korea have developed a groundbreaking AI model called MSI-SEER, which advances the prediction of microsatellite status in cancer. This innovative deep Gaussian process-based Bayesian model analyzes hematoxylin and eosin-stained whole-slide images to predict microsatellite status in gastric and colorectal cancers 1

1

2

Source: Medical Xpress

Understanding Microsatellite Status and Its Importance

Microsatellite status is a crucial indicator of cancer outcomes. Patients with microsatellite instability-high (MSI-H) cancers generally have more promising outcomes compared to those with microsatellite stable tumors. Additionally, tumors deficient in mismatch repair proteins respond well to immune checkpoint inhibitors (ICIs) but not necessarily to chemotherapeutics 1

1

MSI-SEER: A New Approach to Cancer Prediction

The MSI-SEER model addresses shortcomings in previous AI approaches by incorporating uncertainty prediction and quantification. Key features of the model include:

1. Self-assessment of confidence through predictive variance estimation via Monte Carlo dropout
2. Transformation of variance into a Bayesian Confidence Score (BCS)
3. Automatic flagging of high-uncertainty cases for secondary review by human pathologists 1

   1

   2

   2

Validation and Performance

Prof. Jae-Ho Cheong from Yonsei University College of Medicine stated, "We performed extensive validation using multiple large datasets comprising patients from diverse racial backgrounds and found that MSI-SEER achieved state-of-the-art performance with MSI prediction by integrating uncertainty prediction" 1

1

The model demonstrated high accuracy in predicting ICI responsiveness by integrating tumor MSI status and stroma-to-tumor ratio. Furthermore, tile-level predictions provided insights into the contribution of spatial distribution of MSI-H regions in the tumor microenvironment and ICI response 2

2

Source: News-Medical

Clinical Implications and Future Applications

MSI-SEER has the potential for real-world application in prospective cohort surveillance and Phase IV clinical trials. Prof. Cheong emphasized, "The longer-term implication of this study is that it is not about a single specific predictive AI model. Rather, it has a broader implication of how AI algorithms can analyze clinical multi-modal data and create clinically usable models for precision cancer medicine" 2

2

Related Stories

AI analyzes tumor biopsies to predict immunotherapy success in rare cancers, quadrupling survival

ChatGPT Like AI Model Shows Promise in Cancer Treatment Decision-Making

AI Tool SCORPIO Predicts Cancer Immunotherapy Response Using Routine Blood Tests

AI Model Revolutionizes Colorectal Cancer Diagnostics by Detecting Multiple Genetic Markers

AI-Human Collaboration in Cancer Care

The integration of uncertainty prediction in MSI-SEER creates an AI-Human collaboration framework for safer and more reliable clinical environments. By recognizing instances where its predictions carry high uncertainty, the model effectively "knows what it does not know" and defers to human expertise when necessary 1

1

2

Conclusion

The development of MSI-SEER represents a significant advancement in the field of cancer prediction and treatment. By combining state-of-the-art AI technology with uncertainty quantification, this model has the potential to improve cancer diagnosis, treatment decisions, and ultimately, patient outcomes.