1380. Machine Learning-based Estimation of Unconfirmed COVID-19 Cases from a 10,000-Household Survey in Gilgit-Baltistan, Pakistan
Bibliographic record
Abstract
Abstract Background Robust estimates of COVID-19 prevalence during the pandemic are scarce, particularly in settings with limited SARS-CoV-2 testing. Gilgit-Baltistan (GB) is a remote region of Pakistan where healthcare access is limited by underdeveloped facility and road infrastructure. We leveraged a large household survey to describe the burden of confirmed and unconfirmed COVID-19 in GB. Methods We conducted a cross-sectional survey in GB from June–August 2021 during the baseline phase of a cluster randomized trial. Households were randomly selected using a stratified, two-stage sampling design. Data regarding SARS-CoV-2 testing, healthcare worker (HCW) diagnoses without testing, symptoms, and outcomes since March 2020 were self-reported for all household members. “Confirmed/probable” COVID-19 was defined as a positive test, HCW diagnosis of COVID-19, or HCW diagnosis of pneumonia with COVID-19 positive contact. Using machine learning (ML) and bootstrap validation, we developed a symptom-based diagnostic model to differentiate confirmed/probable infections from those with negative SARS-CoV-2 tests (Fig. 1). We applied this model to untested respondents to estimate the total prevalence of COVID-19.Figure 1.Workflow diagram for machine learning analysis. auROC=Area under the receiver operating characteristic curve; CI=Confidence interval; LR=Logistic regression; RF=Random forest; SVM=Support vector machines; XGB=eXtreme Gradient Boosting Results Data were collected from 77924 people in 10264 households. Overall, 314 had confirmed/probable COVID-19, 3263 had negative tests, and 74347 were untested. SARS-CoV-2 testing was less common in females (vs. males; 38 vs. 58 tests per 1000 people) and children (vs. adults; 17 vs. 76 tests per 1000 people). Using an extreme gradient boosting model, area under the receiver operating characteristic curve was 0.92 (95% confidence interval [CI] 0.90–0.93), sensitivity was 0.81 (CI 0.75–0.85), and specificity was 0.88 (CI 0.85–0.90). With this model, total estimated cases were 8–17 times more than the number of individuals with positive tests (Fig. 2). The ratio of estimated to confirmed cases was higher for children (90–213 times) and females (13–25 times).Figure 2.Estimation of probable and possible COVID-19 cases, overall and by age and sex.Confirmed COVID-19 indicates individuals with positive SARS-CoV-2 tests; probable COVID-19 includes HCW diagnoses of COVID-19 and positive predictions from the machine learning analysis; possible COVID-19 includes individuals with positive close contacts or HCW diagnoses of pneumonia. Ratios are depicted as a plausible range between ‘confirmed : probable’ and ‘confirmed : probable + possible’. Conclusion From March 2020–August 2021, the majority of COVID-19 cases in GB went unconfirmed. Women and children were tested less often, perhaps due to preferences in healthcare seeking and perceptions of lower risk of severe illness. Our approach may be used to estimate COVID-19 prevalence in settings with limited testing capacity. Disclosures Shaun Morris, MD, MPH, DTM&H, FRCPC, FAAP, GlaxoSmithKline: Honoraria|JNJ China: Honoraria|Merck: served on ad hoc advisory board|Pfizer: Grant/Research Support|Pfizer: served on ad-hoc advisory board|Sanofi-Pasteur: served on ad-hoc advisory board
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 distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.010 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.001 | 0.002 |
| 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 teacher head, 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".