Performance of Hearing Test Software Applications to Detect Hearing Loss
Bibliographic record
Abstract
Importance: Hearing loss is common and may impact health and quality of life if not properly managed. It is diagnosed following formal audiological assessment, which may not be available or practical. Hearing test software applications (apps) may help identify people who might benefit from audiological assessment, but their diagnostic accuracy has been incompletely studied. Objective: To measure and compare the validity and reliability of 2 commonly recommended apps (hearWHO and SHOEBOX) to detect moderately severe or greater hearing loss. Secondary objectives were to evaluate the apps' ability to detect less severe hearing loss and the diagnostic performance of 2 questionnaires for detecting both severities of hearing loss. Design, Setting, and Participants: This prospective diagnostic accuracy study compared the hearWHO and SHOEBOX apps with a 4-frequency pure-tone average audiological assessment reference standard. All consenting English-speaking patients aged 18 years or older and referred for routine audiological assessment at a publicly funded health center in Calgary, Canada, were included between May 17, 2023, and March 12, 2024. Main Outcome and Measures: The main outcome was the validity and reliability of 4 index tests, including the hearWHO app, SHOEBOX app, Revised Hearing Handicap Inventory-Screening (RHHI-S) questionnaire, and the Single-Item Self-Assessment (SISA) questionnaire, to detect moderate to severe hearing loss. All index test results were compared with an audiological assessment reference standard (hearing loss defined by a better ear hearing threshold of ≥50 dB [more severe denoted as HL50] or ≥20 dB [less severe denoted as HL20]). Test-retest reliability of the 2 apps and C statistics, sensitivity, specificity, and positive and negative predicted values of all index tests were measured. Results: A total of 130 participants were recruited (median [IQR] age, 58 [47-67] years; 82 female [63.1%]). Complete data for each comparison ranged from 123 to 129 participants. The prevalence of HL50 was 16.3% (21 or 130 participants). Neither the hearWHO nor the SHOEBOX app had high test-retest reliability (all κ-values <0.80), with the SHOEBOX having a κ of 0.64 (95% CI, 0.48-0.79) and hearWHO having a κ of 0.32 (95% CI, 0.18-0.46). All C statistics for HL50 were less than 0.80. When testing for HL50, diagnostic performance for both apps was better for the second measurement than the first measurement or the mean. Sensitivity and specificity for the second measurement of SHOEBOX were 0.26 (95% CI, 0.09-0.51) and 1.00 (95% CI, 0.97-1.00), respectively, and for the second measurement of hearWHO, 0.67 (95% CI, 0.43-0.85) and 0.71 (95% CI, 0.62-0.79), respectively. Sensitivity and specificity for the RHHI-S were 0.76 (95% CI, 0.53-0.92) and 0.42 (95% CI, 0.32-0.52), respectively, and for SISA, 0.10 (95% CI, 0.01-0.30) and 0.90 (95% CI, 0.83-0.95), respectively. Using a less stringent diagnostic threshold with SHOEBOX increased sensitivity for HL50 to at least 95% while retaining a specificity of 47% to 54%. Sensitivity and specificity for both apps were higher for HL20. Conclusions and Relevance: These findings suggest that both hearWHO and SHOEBOX have limited test-retest reliability, perhaps because of a learning effect. Both apps may be suitable if a sensitive strategy is desired for identifying people who may benefit from diagnostic audiological assessment, whereas the SHOEBOX app may be preferable if a specific strategy is desired. If neither app is available, the RHHI-S or the SISA could be used depending on whether sensitivity or specificity is desired.
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 | 0.001 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.001 | 0.001 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 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".