Clinical medicine entity identification method and clinical test knowledge mining method

Abstract

The invention discloses a clinical medicine entity identification method and a clinical test knowledge mining method, and the clinical test knowledge mining method comprises the steps of clinical medicine entity identification, clinical test argument mining and clinical test result reasoning. According to the method, on the basis of a BERT pre-model, further pre-training of abstract literatures of RCT is added, a text model in the clinical test field is obtained, entity category information is fused into model prediction, and the effect of a clinical test entity mining task is improved; in the information extraction stage, while a PICO entity is recognized, IC and an evidence sentence containing O are matched, then the logic relation of the IC and the O is judged, and the mask language model architecture of BERT is fully utilized for prediction.

Description

technical field [0001] The invention relates to a knowledge identification method and a mining method, in particular to a clinical medical entity identification method and a clinical trial knowledge mining method. Background technique [0002] Clinical trials refer to medical trials with human bodies (patients or healthy subjects) as objects, aimed at discovering or verifying a certain experimental treatment, including systematic trials of the safety and effectiveness of drugs, devices, vaccines or other treatments . The basis of clinical research is clinical assumptions and clinical needs, and the formulation of clinical assumptions and needs must follow the PICO principle: "P" refers to a specific patient population (target population, population), "I" refers to intervention, " C" refers to the control or another comparable intervention (comparator), and "O" refers to the measurement index (outcome). Based on the PICO principle, the clinical trial protocol design needs t...

AI Technical Summary

Problems solved by technology

[0006] Most of the above work relies on data and algorithms. Most of the mined information is redundant and has low accuracy. It is difficult to provide strong assistance for clinical trial workers, and there is still a long way to go before it can be combined with practical applications.

Moreover, the prediction of clinical trial results ignores the complexity of clinical trials, and the designed relationship classification does not take into account the multiple relationships between interventions and trial results in trials

Images