A Dialogue-based Information Extraction System for Medical Insurance Assessment (2107.05866v1)

Abstract: In the Chinese medical insurance industry, the assessor's role is essential and requires significant efforts to converse with the claimant. This is a highly professional job that involves many parts, such as identifying personal information, collecting related evidence, and making a final insurance report. Due to the coronavirus (COVID-19) pandemic, the previous offline insurance assessment has to be conducted online. However, for the junior assessor often lacking practical experience, it is not easy to quickly handle such a complex online procedure, yet this is important as the insurance company needs to decide how much compensation the claimant should receive based on the assessor's feedback. In order to promote assessors' work efficiency and speed up the overall procedure, in this paper, we propose a dialogue-based information extraction system that integrates advanced NLP technologies for medical insurance assessment. With the assistance of our system, the average time cost of the procedure is reduced from 55 minutes to 35 minutes, and the total human resources cost is saved 30% compared with the previous offline procedure. Until now, the system has already served thousands of online claim cases.

• The paper presents the Intelligent Insurance Assessment System (IIAS), reducing average assessment time from 55 to 35 minutes and cutting costs by 30%.
• It employs advanced NLP techniques, including streaming ASR, BERT, NER/EL, dialogue state tracking, and sentence similarity learning to extract vital information.
• Evaluations demonstrate that the system enhances efficiency and aids less experienced assessors through real-time, dynamic report recommendations.

The paper "A Dialogue-based Information Extraction System for Medical Insurance Assessment" presents an innovative approach to streamline the process of medical insurance evaluation by leveraging advanced NLP technologies. The focus is on aiding assessors, particularly those who are less experienced, in handling complex insurance claims efficiently through an intelligent system called the Intelligent Insurance Assessment System (IIAS).

Key Contributions and Features:

1. System Overview: The IIAS is designed to facilitate online insurance assessment processes, which became crucial during the COVID-19 pandemic due to restrictions on in-person interactions. The system reduces the average assessment time significantly from 55 to 35 minutes and cuts human resource costs by 30%, benefiting both insurance companies and claimants.
2. Technological Integration: The system integrates several state-of-the-art NLP techniques:
   • Streaming Automatic Speech Recognition (ASR): Converts spoken interactions into text, serving as a foundational component for further processing.
   • Pre-trained LLMs: Employs models like BERT for understanding and processing language efficiently.
   • Named Entity Recognition/Linking (NER/EL): Identifies and categorizes entities (e.g., addresses, hospital names) within the dialogue.
   • Dialogue State Tracking (DST): Helps in maintaining the contextual state of the conversation to filter out irrelevant information, improving the accuracy of information extraction.
   • Sentence Similarity Learning: Assists in topic detection during conversations to organize dialogue into coherent segments.
3. Applications and Usability: IIAS enhances assessors' productivity by automatically extracting information and suggesting relevant data while filling out insurance reports. The system displays keywords in real-time on a dashboard and recommends content dynamically, aiding report compilation.
4. Evaluation and Performance: The paper details both offline and online assessments of the system. Offline evaluations utilize a manually constructed dataset, showing favorable results in extracting and suggesting relevant information. Online evaluations confirm substantial improvements in the efficiency of the insurance assessment process.
5. Challenges and Future Directions: The system identifies challenges such as dependence on ASR accuracy, difficulties in handling reasoning-based questions, and cumulative errors from multi-model approaches. Future work includes optimizing ASR and integrating more advanced reasoning capabilities into the system.

Conclusion:

The paper highlights the potential of integrating dialogue-based information extraction systems in professional settings like medical insurance assessments. By providing significant time and cost efficiencies and lowering entry barriers for less experienced assessors, IIAS demonstrates a practical application of NLP technologies that could expand to other domains requiring dialogue-centric data extraction and report generation.