PhthisisBioMed Artificial Medical Intelligence: Software for Automated Analysis of Digital Chest X-ray/Fluorograms
Bibliographic record
Abstract
The scope of diagnostic medical examinations increases from year to year causing a reasonable desire to develop and implement new technologies to diagnostics and medical data analysis. Artificial intelligence (AI) algorithms became one of the most promising solutions to this problem and proved themselves in the course of mass practical application. During the three-year Moscow experiment started in 2020, the possibility was achieved to develop methodologies of AI use and to successfully implement it into the regional level healthcare system. In this article, the authors share their experience in developing a medical AI service using the example of PhthisisBioMed AI service and the results of its application in real clinical activities environment. This AI service has shown its quality and reliability confirmed by technological monitoring. Clinical trials of PhthisisBioMed AI service were conducted on a specially prepared verified data set (n=1536) considering epidemiological indicators of the thoracic organs major diseases prevalence. The mean sensitivity of the service was 0.975 (95% CI: 0.966-0.984). PhthisisBioMed medical AI service is registered as a medical device (medical device registration certificate No.RZN 2022/17406 dated May 31, 2022) and is actively used in the Russian Federation as a diagnostic tool to reduce the burden on radiologists and to accelerate the process of medical report obtaining.
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 | 0.007 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.003 | 0.002 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.001 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.012 | 0.006 |
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".