Sahel medical intelligence using ChatGPT. One more tool but not the definitive solution

Abstract

Abstract Purpose The Sahel is a region in conflict where the development of medical intelligence (MEDINT) presents various complications. The ChatGPT application is a tool based on artificial intelligence that can be useful in some phases of the intelligence cycle. Methods We designed a series of 20 prompts oriented to the different phases of the MEDINT cycle and the answers were analyzed by 4 experts using a Likert scale according to criteria of relevance (range 1– completely irrelevant to 5 – completely relevant) and coherence (range 1– completely incoherence to 5 – completely coherence). Scores were summarized with descriptive statistics. Results Among the 160 evaluations generated after the analysis by the experts, the median accuracy score was 3 (mean 3.25, SD 0.99). In direction (mean 4.08, SD 0.14) and elaboration (mean 3.92, SD 0.67), the application shows above-average results in both criteria. In the rest of the phases, the results are consistent, but not relevant. Experts discreetly consider coherence (mean 2.5, SD 0.33) Conclusions: 1) ChatGPT can be useful as a guide when creating a MEDINT plan for a region. 2) In the rest of the phases, the application itself recognizes its limitations due to not being able to access information in real time or analyze new data. Artificial intelligence systems based on large language models, can be configured for specific topics and access to restricted data. Therefore, we recommend the advanced study of use of these tools in the field of medical intelligence.