Specialty
Sepsis Prediction
Sepsis-prediction models have a documented history of high-profile real-world generalizability failures. Appraisals here weigh prospective validation and calibration in the deployment setting, not just retrospective discrimination.
Appraised tools compared
- Tool
- AI and Healthcare: What Real Clinical Deployments Actually Look Like
- FDA status
- Not reported
- External validation
- Not reported
- Manufacturer
- —
Artificial Intelligence in the Medical Field: How Clinical Deployments Actually Work
- Tool
- Artificial Intelligence in the Medical Field: How Clinical Deployments Actually Work
- FDA status
- Not reported
- External validation
- Not reported
- Manufacturer
- —
neurology
AI for Parkinson Disease Diagnosis: What the Evidence Shows
- Tool
- AIDP
- FDA status
- FDA De Novo classification DEN240071
- External validation
- Not independently validated
- Key metric
- AUROC >96%
Neurology
AI-powered Parkinson's support groups have almost no published evidence
- Tool
- Patrika
- FDA status
- Not FDA-cleared
- External validation
- No external validation
- Key metric
- 99% intent-identification accuracy
Sports Medicine
AI Sports Injury Recovery Prediction Lacks Validated Clinical Evidence
- Tool
- Return-to-sport ML models
- FDA status
- No FDA clearance identified for this indication
- External validation
- Limited external validation
- Key metric
- AUC up to 0.96
Cardiology
How Much Does the Flu Shot Really Cut Atrial Fibrillation Risk?
- Tool
- Influenza vaccination
- FDA status
- Not applicable
- External validation
- No
- Key metric
- HR 0.80
Neurology
How Foundations Are Powering Parkinson's AI Research
- Tool
- PPMI, GP2, PD GENEration
- FDA status
- No FDA clearance (non-device)
- External validation
- Not established
- Manufacturer
- Michael J. Fox Foundation, Parkinson's Foundation, ASAP, APDA
Recent tracker activity
Open full trackerVocabulary for this specialty
Commonly needed to read sepsis prediction evidence: calibration, prospective validation, risk of bias.
Browse the full glossary