ML Record Evaluation via Text Block Segmentation

The patent segments the medical record into multiple blocks of text, each representing a distinct section or paragraph. This segmentation allows the system to process and evaluate each block independently, identifying relevant information while filtering out noise. The segmentation principle resolves the contradiction by enabling automated processing of individual blocks while maintaining overall evaluation accuracy through cumulative analysis of all blocks.

The patent extracts relevant features and information from each text block, separating the signal (clinically relevant information) from the noise (irrelevant or redundant text). This extraction process involves identifying key entities, relationships, and patterns that are critical for accurate medical record evaluation. By taking out only the essential information, the system achieves both automation and high accuracy.

The patent performs preliminary processing of the medical record by segmenting it into blocks and pre-identifying potentially relevant sections before full evaluation. This preliminary action includes basic text processing, block segmentation, and initial relevance assessment. By performing these actions beforehand, the system reduces the computational burden during the main evaluation phase, maintaining comprehensiveness while reducing processing time.

The patent applies partial processing to each text block, focusing computational resources on the most relevant blocks while using lighter processing for less critical sections. This approach involves evaluating blocks at different levels of detail, applying more rigorous analysis only where necessary. This partial action strategy maintains overall evaluation accuracy while significantly reducing total processing time and computational resource requirements.