Automatic Patient Search Using Bernoulli Model

Objective: Develop algorithms to automatically identify qualified patients for breast cancer clinical trials from free-text medical reports. Design: The Bernoulli model was trained to search for a qualified patient based on criterion. Measurement: The performance of the Bernoulli model was evaluated by the Precision-Recall curve and F-score. Results: The Single-word Bernoulli model trained in a two-class mode has greater performance than the model trained in a one-class mode. The performance of the model was also compared with some other techniques. Conclusions: The Bernoulli model method is easier to implement and performs better than several competing techniques.