Artificial Intelligence–Powered Training Database for Clinical Thinking: App Development Study
Bibliographic record
Abstract
Background: With the development of artificial intelligence (AI), medicine has entered the era of intelligent medicine, and various aspects, such as medical education and talent cultivation, are also being redefined. The cultivation of clinical thinking abilities poses a formidable challenge even for seasoned clinical educators, as offline training modalities often fall short in bridging the divide between current practice and the desired ideal. Consequently, there arises an imperative need for the expeditious development of a web-based database, tailored to empower physicians in their quest to learn and hone their clinical reasoning skills. Objective: This study aimed to introduce an app named "XueYiKu," which includes consultations, physical examinations, auxiliary examinations, and diagnosis, incorporating AI and actual complete hospital medical records to build an online-learning platform using human-computer interaction. Methods: The "XueYiKu" app was designed as a contactless, self-service, trial-and-error system application based on actual complete hospital medical records and natural language processing technology to comprehensively assess the "clinical competence" of residents at different stages. Case extraction was performed at a hospital's case data center, and the best-matching cases were differentiated through natural language processing, word segmentation, synonym conversion, and sorting. More than 400 teaching cases covering 65 kinds of diseases were released for students to learn, and the subjects covered internal medicine, surgery, gynecology and obstetrics, and pediatrics. The difficulty of learning cases was divided into four levels in ascending order. Moreover, the learning and teaching effects were evaluated using 6 dimensions covering systematicness, agility, logic, knowledge expansion, multidimensional evaluation indicators, and preciseness. Results: From the app's first launch on the Android platform in May 2019 to the last version updated in May 2023, the total number of teacher and student users was 6209 and 1180, respectively. The top 3 subjects most frequently learned were respirology (n=606, 24.1%), general surgery (n=506, 20.1%), and urinary surgery (n=390, 15.5%). For diseases, pneumonia was the most frequently learned, followed by cholecystolithiasis (n=216, 14.1%), benign prostate hyperplasia (n=196, 12.8%), and bladder tumor (n=193, 12.6%). Among 479 students, roughly a third (n=168, 35.1%) scored in the 60 to 80 range, and half of them scored over 80 points (n=238, 49.7%). The app enabled medical students' learning to become more active and self-motivated, with a variety of formats, and provided real-time feedback through assessments on the platform. The learning effect was satisfactory overall and provided important precedence for establishing scientific models and methods for assessing clinical thinking skills in the future. Conclusions: The integration of AI and medical education will undoubtedly assist in the restructuring of education processes; promote the evolution of the education ecosystem; and provide new convenient ways for independent learning, interactive communication, and educational resource sharing.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.002 | 0.009 |
| Meta-epidemiology (narrow) | 0.001 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.002 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.004 | 0.001 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".