Prediction of Obstructive Sleep Apnea With AI-based Facial Photograph Analysis
Bibliographic record
Abstract
Abstract Background: Upper airway/craniofacial anatomy is an important contributing factor in the pathogenesis of Obstructive Sleep Apnea (OSA). Aim: This study aimed to evaluate whether facial landmark coordinates and craniofacial morphometric parameters obtained from facial photographs using a smartphone can predict OSA. Methods: We consecutively recruited patients who underwent in-lab polysomnography. A frontal facial photograph of each patient was captured using iPhone 12 (Apple, Inc., CA, USA). The coordinates of 11 facial landmarks in the photograph were determined by a feature point recognition model developed using artificial intelligence (AI). The coordinates of 68 facial landmarks were determined by a feature point recognition model developed using artificial intelligence (AI) and the representative 11 landmarks among them were chosen for further analysis. (Figure 1) At all 55 combinations of two of the 11 points, lines were drawn and lengths of them were measured. In addition, at all 1485 combinations of two of these 55 lines, ratios of length of one to another line were calculated. In total, 1,540 variables (length of 55 lines and 1,485 ratios of length of two lines) were calculated. The correlation of these variables with obstructive apnea hypopnea index (OAHI) were evaluated. Result: The photographs from 124 patients (Men 63, 49±13 years old) were analyzed. The ratio of the length of the line connecting the right edge of the ocular fissure with the edge of the ipsilateral nasal wing to the length of the line crossing the face obliquely was associated with OAHI independently from the confounding factors (β=0.38, p=2.86×10-5). (Figure 2) In addition, the comparison of two prediction models for patients with OAHI>15/h ((1) the model made by age, sex and BMI vs. (2) the model made by age, sex, BMI and the parameter from the facial photograph) demonstrated that the area under curve (AUC) of receiver operating characteristic curve was significantly greater in the model that included the photographic parameter (AUC: 0.74 vs. 0.82, p=0.02) Conclusion: AI analysis of craniofacial morphology from a single facial photograph can be useful in predicting the presence of OSA.
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 | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
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".