Leveraging Symptom Search Data to Understand Disparities in US Mental Health Care: Demographic Analysis of Search Engine Trace Data
Bibliographic record
Abstract
Background In the United States, 1 out of every 3 people lives in a mental health professional shortage area. Shortage areas tend to be rural, have higher levels of poverty, and have poor mental health outcomes. Previous work has demonstrated that these poor outcomes may arise from interactions between a lack of resources and lack of recognition of mental illness by medical professionals. Objective We aimed to understand the differences in how people in shortage and nonshortage areas search for information about mental health on the web. Methods We analyzed search engine log data related to health from 2017-2021 and examined the differences in mental health search behavior between shortage and nonshortage areas. We analyzed several axes of difference, including shortage versus nonshortage comparisons, urban versus rural comparisons, and temporal comparisons. Results We found specific differences in search behavior between shortage and nonshortage areas. In shortage areas, broader and more general mental health symptom categories, namely anxiety (mean 2.03%, SD 0.44%), depression (mean 1.15%, SD 0.27%), fatigue (mean 1.21%, SD 0.28%), and headache (mean 1.03%, SD 0.23%), were searched significantly more often (Q<.0003). In contrast, specific symptom categories and mental health disorders such as binge eating (mean 0.02%, SD 0.02%), psychosis (mean 0.37%, SD 0.06%), and attention-deficit/hyperactivity disorder (mean 0.77%, SD 0.10%) were searched significantly more often (Q<.0009) in nonshortage areas. Although suicide rates are consistently known to be higher in shortage and rural areas, we see that the rates of suicide-related searching are lower in shortage areas (mean 0.05%, SD 0.04%) than in nonshortage areas (mean 0.10%, SD 0.03%; Q<.0003), more so when a shortage area is rural (mean 0.024%, SD 0.029%; Q<2 × 10–12). Conclusions This study demonstrates differences in how people from geographically marginalized groups search on the web for mental health. One main implication of this work is the influence that search engine ranking algorithms and interface design might have on the kinds of resources that individuals use when in distress. Our results support the idea that search engine algorithm designers should be conscientious of the role that structural factors play in expressions of distress and they should attempt to design search engine algorithms and interfaces to close gaps in care.
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 | 0.002 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.001 | 0.003 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.001 | 0.002 |
| Research integrity | 0.000 | 0.001 |
| 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".