What we track

Texas Health Monitor is a public social-listening dashboard built on the Socialhose Public API. It aggregates social and news mentions across 9 campaigns, 412 keywords, and 161 live searches running on six platforms: X (Twitter), Reddit, Facebook, Instagram, TikTok, and LinkedIn.

Coverage is organized into two complementary dimensions:

• Statewide verticals cover broad Texas-wide conversation within one healthcare segment (Hospitals & Health Systems, Clinics & Physician Practices, Pharmacies & Retail Health, Drugs / Prescriptions / Therapeutics).
• Metro cross-verticals cover all healthcare conversation localized to one major market (Houston, Dallas–Fort Worth, Austin, San Antonio, El Paso).

How we collect it

Each campaign uses live searches, which are platform-specific scheduled queries that collect matching documents per run. Searches are tuned per platform: Boolean AND/OR queries on X, Reddit, Facebook, TikTok and LinkedIn; single bare hashtags on Instagram (the only format that platform accepts).

The top 5 hospital systems (Memorial Hermann, Houston Methodist, MD Anderson, UT Southwestern, Parkland) and the top 5 drugs (Ozempic, Wegovy, Mounjaro, Zepbound, Adderall) each get dedicated single-entity feeds on X and TikTok. This lets the dashboard show per-brand sentiment and volume without slicing it out of a larger OR-chain query.

Cadence & freshness

Most searches run on a 5-day standard cadence. Ten high-volume topics run daily (high-frequency tier): the GLP-1 cluster and dedicated drug feeds on X and TikTok, the ADHD-shortage cluster on X, the TX hospital and pharmacy brand-broad clusters on X, and dedicated MD Anderson on X. No searches run on the hourly crisis tier. This is a research program, not a real-time alerting pipeline, and incident-themed topics are low-volume between events.

The dashboard captures the most relevant mentions for each narrow sub-topic, not a comprehensive census. High-volume topics with saturated feeds may miss long-tail posts.

Geographic scoping

Every search is scoped to Texas through one of six mechanisms:

• Geo clause (104 searches): explicit AND-clause with TX/Houston/Dallas/Austin/etc. terms.
• TX brand (16 searches): the query references a Texas-only or Texas-anchored brand (e.g. Memorial Hermann, HEB Pharmacy), so the brand itself is the geo signal.
• Subreddit (9 searches): Reddit query uses a TX city subreddit filter.
• Hashtag brand (24 Instagram searches): single bare hashtag is a TX-only brand (MemorialHermann, MDAnderson, etc.).
• Hashtag geo (8 Instagram searches): single bare hashtag is a niche TX healthcare-geo tag (HoustonHealth, TMC, KeepAustinHealthy, etc.).

Instagram has a hard constraint: live searches accept exactly one bare hashtag, with no Boolean operators and no geo AND-clause. This means Instagram coverage is limited to brand-specific or geo-specific hashtags. The Drugs campaign deliberately has zero Instagram searches because drug-name hashtags are global and cannot be scoped to Texas under this rule.

Languages

English and Spanish keywords are woven across every campaign. San Antonio and El Paso carry dedicated Spanish-language platform searches because they are the most Spanish-dominant healthcare markets in Texas. The El Paso campaign also includes a borderland search that captures Ciudad Juárez pharmacy and care conversation, which is signal that typical Texas-only listening misses.

Sentiment & classification

Mentions are machine-classified by Socialhose into positive, neutral, or negative sentiment with a confidence score. The dashboard shows sampled sentiment for entities with low mention counts; exact sentiment distributions are computed by facet when volume is high enough. Sentiment is directional, not diagnostic. A "negative" hospital mention could be a patient complaint, a news report about a cyberattack, or a union grievance.

Known limitations

• Platform gaps. Instagram cannot carry generic drug or geo-only hashtags under the single-hashtag rule, so Drugs has no Instagram coverage. Some platforms are skipped per campaign where they add no signal.
• Geo inference is imperfect. A TX brand query (e.g. "Memorial Hermann") will pick up out-of-state patient travel posts. Geo-clause queries miss posts that omit location terms. The dashboard biases toward captured Texas conversation, not a complete census.

How to read the numbers

Total mentions is the count of collected posts matching the entity’s search criteria. Last 7 days sums the trailing 7 days of daily volume; the delta compares that sum to the prior 7 days. Positive / Neutral / Negative shares are percentages of the sampled or exact population. Platform mix shows where the conversation happened in the collected sample.

Daily digests

Each campaign has a daily mailing-list digest. Within the daily run, keyword matches or sentiment thresholds can promote that day’s email to a richer template, but the cadence stays daily. You can subscribe to digests here.