Developing a Multisensor-Based Machine Learning Technology (Aidar Decompensation Index) for Real-Time Automated Detection of Post–COVID-19 Condition: Protocol for an Observational Study
Bibliographic record
Abstract
BACKGROUND: Post-COVID-19 condition is emerging as a new epidemic, characterized by the persistence of COVID-19 symptoms beyond 3 months, and is anticipated to substantially alter the lives of millions of people globally. Patients with severe episodes of COVID-19 are significantly more likely to be hospitalized in the following months. The pathophysiological mechanisms for delayed complications are still poorly understood, with a dissociation seen between ongoing symptoms and objective measures of cardiopulmonary health. COVID-19 is anticipated to alter the long-term trajectory of many chronic cardiovascular and pulmonary diseases, which are common among those at risk of severe disease. OBJECTIVE: This study aims to use a single, integrated device-MouthLab, which measures 10 vital health parameters in 60 seconds-and a cloud-based proprietary analytics engine to develop and validate the Aidar Decompensation Index (AIDI), to predict decompensation in health among patients who previously had severe COVID-19. METHODS: Overall, 200 participants will be enrolled. Inclusion criteria are patients in the US Department of Veterans Affairs health care system; "severe" COVID-19 infection during the acute phase, defined as requiring hospitalization, within 3-6 months before enrollment; aged ≥18 years; and having 1 of 6 prespecified chronic conditions. All participants will be instructed to use the MouthLab device to capture daily physiological data and complete monthly symptom surveys. Structured data collection tables will be developed to extract the clinical characteristics of those who experience decompensation events (DEs). The performance of the AIDI will depend on the magnitude of difference in physiological signals between those experiencing DEs and those who do not, as well as the time until a DE (ie, the closer to the event, the easier the prediction). Information about demographics, symptoms (Medical Research Council Dyspnea Scale and Post-COVID-19 Functional Status Scale), comorbidities, and other clinical characteristics will be tagged and added to the biomarker data. The resultant predicted probability of decompensation will be translated into the AIDI, where there will be a linear relationship between the risk score and the AIDI. To improve prediction accuracy, data may be stratified based on biological sex, race, ethnicity, or underlying clinical characteristics into subgroups to determine if there are differences in performance and detection lead times. Using appropriate algorithmic techniques, the study expects the model to have a sensitivity of >80% and a positive predicted value of >70%. RESULTS: Recruitment began in January 2023, and at the time of manuscript submission, 204 patients have been enrolled. Publication of the complete results and data from the study is expected in 2025. CONCLUSIONS: The focus on identifying predictor variables using a combination of biosensor-derived physiological features should enable the capture of heterogeneous characteristics of complications related to post-COVID-19 condition across diverse populations. TRIAL REGISTRATION: ClinicalTrials.gov NCT05220306; https://clinicaltrials.gov/study/NCT05220306.
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 | 0.024 | 0.023 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.002 | 0.003 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.003 | 0.002 |
| Scholarly communication | 0.002 | 0.002 |
| Open science | 0.002 | 0.002 |
| Research integrity | 0.002 | 0.003 |
| Insufficient payload (model declined to judge) | 0.027 | 0.007 |
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".