NLP in insurance medicine / Research into the support of insurance physicians in their work using Natural Language Processing (NLP)

Lausberg, (Christine) MA BN, Wit de, (Mariska) PhD, Syurina, (Elena) PhD, Kam de, (Daniël) PhD, Boer de, (Angela) MD, PhD, Anema, (Han) MD, PhD

About the junior researcher

My name is Christine Lausberg. In December 2024, I started my PhD research at the KCVG, Amsterdam UMC/ VU location. Besides my work as a PhD candidate at the KCVG, I work as a social medical nurse at UWV in Eindhoven.

Background of the research project

Currently, administrative burdens have a significant impact on the workload and job satisfaction of medical professionals working within insurance medicine. These burdens are increasing due to rising staff shortages and a growing demand for social medical assessments. For clients, it is important that they receive a timely, high-quality, and, above all, correct assessment, but also clarity regarding their income and return to work options. The use of Artificial Intelligence (AI) can be helpful in this regard. Primarily Natural Language Processing (NLP) and NLP-based support could be used to reduce the administrative burden on healthcare professionals. NLP uses various techniques to analyze spoken or written text and generate summaries, transcripts, or reports. The existing technology can be deployed on various fronts and is already being used successfully in multiple sectors.

Research objectives

The goal of this study is to investigate how NLP can be used for administrative support within the insurance medical practice, but also to identify the preconditions for acceptance and thorough implementation. We expect that administrative support through NLP will contribute to the overall quality and effectiveness of the social medical assessment. For example, NLP can reduce practice variation among professionals in reporting. Additionally, we expect that NLP can help promote efficiency by reducing the administrative burden for the insurance physician. This may potentially reduce waiting times for clients through faster processing of applications.

Method

The research consists of different sub-projects. It starts with an inventory of existing NLP systems aimed at administrative support within the curative sector, occupational health, and private insurance medicine. (Medical-) professionals working in the aforementioned sectors will be interviewed to get a picture of their experiences with using NLP. Factors that hinder and facilitate acceptance and implementation of an NLP-based tool will also be examined.

Subsequently, a validation study will investigate the extent to which NLP can be used to summarize data and reports in insurance medical practice and whether this output is of sufficient quality, accurate, and without bias. In a later phase of this project, the effects and cost-effectiveness of administrative support for insurance physicians in practice will be examined.

Product and implications for practice

By gaining insight into existing tools and the factors that influence their acceptance, as well as by focusing on thorough implementation, we hope that in the future NLP-based support can help insurance physicians to:

• Improve the quality of the (medical) reports.
• Reduce the administrative burden for the insurance physician, and other collegues (e.g. social medical nurses and medical secretaries).
• Reduce mutual differences in practice variation and the quality of assessments and reasoning among insurance physicians.
• Work more efficiently when assessing clients by saving time during staff shortages.
• Reduce waiting times for the client.

Contactinformation

Email: c.lausberg@amsterdamumc.nl
Update: 7-5-2025