Skip to content
HALEMedical and Data Sciences

Capabilities

Applied research on the data agencies already hold.

Three capabilities, one field. We design the study, run the analysis, and say what the result does and does not support — on existing clinical, claims, and public-health data, under one standard: pre-specified methods, reproducible analysis, and evidence a review board can read.

The consoles on this page are illustrative product concepts, not screenshots of delivered systems.

NAICS 541715· Primary

Life & Health Sciences R&D

Study design and analysis on data agencies already hold. From forecasting an outbreak to measuring whether a treatment worked: retrospective and observational research, clinical outcomes, epidemiological modeling, and health services research for government health programs. Scenario models that quantify what an intervention buys, time-to-event analysis that shows who benefits, and statistics rigorous enough to publish and reproducible enough to audit.

Hale MDS · Population health

SYNDROMIC · WK 36

Surge probability

78%

12wk · P(≥28/100k) · peak wk 43

Projected peak

31.4

per 100k · wk 43 · CI 26–37

Lead time

6 wk

median crossing · wk 42

Respiratory admissions per 100k — 12-week surge outlook

102030MAYJULSEPOCTNOVFORECASTSURGE THRESHOLD · 28crossing · wk 42P(SURGE) BY WEEKWeek 37 — 1% chance admissions exceed 28 per 100k (baseline)1Week 38 — 3% chance admissions exceed 28 per 100k (baseline)3Week 39 — 8% chance admissions exceed 28 per 100k (baseline)8Week 40 — 19% chance admissions exceed 28 per 100k (baseline)19Week 41 — 38% chance admissions exceed 28 per 100k (baseline)38Week 42 — 61% chance admissions exceed 28 per 100k (baseline)61Week 43 — 78% chance admissions exceed 28 per 100k (baseline)78Week 44 — 74% chance admissions exceed 28 per 100k (baseline)74Week 45 — 63% chance admissions exceed 28 per 100k (baseline)63Week 46 — 49% chance admissions exceed 28 per 100k (baseline)49Week 47 — 35% chance admissions exceed 28 per 100k (baseline)35Week 48 — 24% chance admissions exceed 28 per 100k (baseline)24%
ObservedMedian forecast80% band95% bandSurge thresholdP(surge) 0–100%
Surge-forecast console

Hale MDS · Geo surveillance

TAMPA BAY · WK 30 · PER 100K

Choropleth map of the Tampa Bay region — real Census ZIP-code boundaries for Hillsborough, Pinellas, and southern Pasco counties — shaded light to dark blue by weekly syndromic rate per 100,000. A three-ZIP cluster in East Tampa is flagged in amber and under review; a milder rise shows in south St. Petersburg. Illustrative rates over real geography.

Low → highFlagged cluster122 ZCTAs · real boundaries, illustrative rates
ZIP-level surveillance, Tampa Bay

Hale MDS · Outcomes research

TIME-TO-EVENT · 24-MO FOLLOW-UP

Time to readmission after discharge — % remaining event-free

255075100%06121824 moCTP · 53%UC · 27%NO. AT RISKCTP142103847351UC13987684020
Care-transition programme (CTP)Usual care (UC)95% CI bandCensored

Adjusted hazard ratios — Cox proportional-hazards model

Filled marker · CI excludes 1.0

0.512HR 1.0◂ LOWER RISKHIGHER RISK ▸HR (95% CI)Care-transition programme (vs usual care) — adjusted hazard ratio 0.55 (0.40–0.76), CI excludes 1.0Care-transition programmevs usual care0.55 (0.40–0.76)Prior hospitalization (≥1 in prior year) — adjusted hazard ratio 1.68 (1.21–2.33), CI excludes 1.0Prior hospitalization≥1 in prior year1.68 (1.21–2.33)Medication adherence ≥80% (first 6 months) — adjusted hazard ratio 0.61 (0.44–0.85), CI excludes 1.0Medication adherence ≥80%first 6 months0.61 (0.44–0.85)Comorbidity burden (≥2 conditions at baseline) — adjusted hazard ratio 1.42 (1.02–1.98), CI excludes 1.0Comorbidity burden≥2 conditions at baseline1.42 (1.02–1.98)Age under 30 (at index admission) — adjusted hazard ratio 0.86 (0.62–1.19)Age under 30at index admission0.86 (0.62–1.19)Stable housing (at discharge) — adjusted hazard ratio 0.78 (0.55–1.10)Stable housingat discharge0.78 (0.55–1.10)
Outcomes console — time-to-event and adjusted hazards

What we deliver

  • Retrospective and observational study design on existing data
  • Clinical outcomes and comparative-effectiveness research
  • Epidemiological modeling, surveillance and geospatial analysis
  • Health services research: access, utilization and cost
  • Time-to-event analysis and hazard estimation
  • Publication-grade statistical analysis, reproducible end to end

NAICS 541714

Applied Analytic Methods

Methods that make secondary data answer a question defensibly — phenotyping, confounding control, bias analysis, and privacy-preserving analytics. Data collected for care and payment was not collected for the question in front of you; the method is what closes that gap, and it is written down before the first query runs. Every workflow is reproducible end to end, because research that can't be rerun can't be trusted.

Hale MDS · Cohort comparison

IPTW · N = 2,418 vs 2,391

Standardized mean differences — before and after weighting

Hollow · before   Filled · after

00.10.20.30.40.5SMD 0.1|STANDARDIZED MEAN DIFFERENCE| ▸Age at index admission — SMD 0.42 before weighting, 0.03 afterAge at index admissionSex — SMD 0.11 before weighting, 0.02 afterSexComorbidity count — SMD 0.36 before weighting, 0.05 afterComorbidity countAdmissions, prior 12 months — SMD 0.45 before weighting, 0.06 afterAdmissions, prior 12 monthsOutpatient visits, prior 12 months — SMD 0.29 before weighting, 0.04 afterOutpatient visits, prior 12 monthsMedication count at discharge — SMD 0.24 before weighting, 0.03 afterMedication count at dischargeIndex year — SMD 0.08 before weighting, 0.01 afterIndex yearRurality of residence — SMD 0.19 before weighting, 0.04 afterRurality of residenceCoverage type — SMD 0.15 before weighting, 0.02 afterCoverage typeAbnormal baseline lab — SMD 0.27 before weighting, 0.05 afterAbnormal baseline lab

Effect on 30-day readmission — care-transition programme vs usual care

Risk difference · 95% CI

Crude
−7.8 pp (−9.6 pp to −6.0 pp)
Weighted
−3.2 pp (−5.1 pp to −1.3 pp)

The crude comparison credits the programme with more than it did: it enrolled the sicker patients. Balancing the cohorts is what makes the two numbers comparable — and the adjusted interval still excludes zero.

Cohort comparison — covariate balance before and after weighting

What we deliver

  • Cohort and case definition on EHR and claims data
  • Confounding control: propensity methods, weighting and matching
  • Bias analysis and sensitivity analysis
  • Missing-data methods and measurement-error assessment
  • Privacy-preserving analytics and de-identification
  • Methods documentation a reviewer can audit

NAICS 541690

Scientific & Technical Consulting

Physician-led advisory to federal health programs: framing the clinical question, designing the protocol that can answer it, and serving as principal investigator or subject-matter expert on the work that follows. The judgment being sold is clinical — what an outcome measure actually captures, which confounders matter in a population, and what a result does and does not license a programme to conclude.

What we deliver

  • Clinical advisory and protocol design for federal health research
  • Principal investigator and subject-matter expert support
  • Outcome-measure selection and endpoint definition
  • Study feasibility, sample-size and power assessment
  • Independent review of research designs and analytic plans
  • Translation of research findings into programme decisions

Reproducible analysis pipelines

Analysis that can be rerun is analysis that can be trusted. Versioned data and code, engineered pipelines with full lineage, documented decisions, and interfaces designed to Section 508 from the first sketch — so a reviewer can regenerate every figure from source and see how it was made.

What we deliver

  • Engineered data and analysis pipelines with full lineage
  • Versioned data and code; every figure regenerates from source
  • Analysis plans written before the first query runs
  • Accessible interfaces designed to Section 508 / WCAG 2.2 AA
  • API design and modernization of legacy service layers
  • Automated testing and documentation as first-class deliverables

Secure research environments

Health data deserves infrastructure built for it. We design the platforms that ingest, process, and serve clinical and claims data at scale — and the research enclaves where sensitive datasets can be analyzed without leaving governed boundaries. Accreditation is treated as an engineering problem: controls as code, evidence generated automatically, environments reproducible from day one.

Hale MDS · Platform operations

US-EAST · ENCLAVE 03

Uptime · 90d

99.996%

SLO 99.95 · enclave fleet

API p95

38 ms

FHIR R4 · trailing 24h

Ingest lag

1.8 s

event → queryable · p95

Error budget

72%

remaining · 30d window

Streaming ingest — trailing 4h, all pipelines

Σ 32.5k ev/s

102030-4H-3H-2H-1HNOWk/sHL7v2 12.8Claims 7.8FHIR R4 8.9Registries 3.0

HL7v2 ADT/ORU · X12 837 claims · FHIR R4 bulk · registry extracts — k events/s

Query latency — p95 by hour of day

FHIR APIFHIR API · 00:00 — p95 13 msFHIR API · 01:00 — p95 12 msFHIR API · 02:00 — p95 13 msFHIR API · 03:00 — p95 12 msFHIR API · 04:00 — p95 11 msFHIR API · 05:00 — p95 12 msFHIR API · 06:00 — p95 13 msFHIR API · 07:00 — p95 14 msFHIR API · 08:00 — p95 16 msFHIR API · 09:00 — p95 19 msFHIR API · 10:00 — p95 22 msFHIR API · 11:00 — p95 25 msFHIR API · 12:00 — p95 32 msFHIR API · 13:00 — p95 30 msFHIR API · 14:00 — p95 31 msFHIR API · 15:00 — p95 26 msFHIR API · 16:00 — p95 23 msFHIR API · 17:00 — p95 18 msFHIR API · 18:00 — p95 15 msFHIR API · 19:00 — p95 13 msFHIR API · 20:00 — p95 13 msFHIR API · 21:00 — p95 13 msFHIR API · 22:00 — p95 13 msFHIR API · 23:00 — p95 12 msQuery engQuery eng · 00:00 — p95 16 msQuery eng · 01:00 — p95 16 msQuery eng · 02:00 — p95 16 msQuery eng · 03:00 — p95 16 msQuery eng · 04:00 — p95 16 msQuery eng · 05:00 — p95 18 msQuery eng · 06:00 — p95 16 msQuery eng · 07:00 — p95 17 msQuery eng · 08:00 — p95 20 msQuery eng · 09:00 — p95 23 msQuery eng · 10:00 — p95 25 msQuery eng · 11:00 — p95 35 msQuery eng · 12:00 — p95 38 msQuery eng · 13:00 — p95 45 msQuery eng · 14:00 — p95 42 msQuery eng · 15:00 — p95 42 msQuery eng · 16:00 — p95 41 msQuery eng · 17:00 — p95 32 msQuery eng · 18:00 — p95 26 msQuery eng · 19:00 — p95 23 msQuery eng · 20:00 — p95 18 msQuery eng · 21:00 — p95 17 msQuery eng · 22:00 — p95 17 msQuery eng · 23:00 — p95 16 msTerminologyTerminology · 00:00 — p95 6 msTerminology · 01:00 — p95 6 msTerminology · 02:00 — p95 6 msTerminology · 03:00 — p95 6 msTerminology · 04:00 — p95 6 msTerminology · 05:00 — p95 6 msTerminology · 06:00 — p95 7 msTerminology · 07:00 — p95 9 msTerminology · 08:00 — p95 11 msTerminology · 09:00 — p95 14 msTerminology · 10:00 — p95 15 msTerminology · 11:00 — p95 17 msTerminology · 12:00 — p95 16 msTerminology · 13:00 — p95 15 msTerminology · 14:00 — p95 12 msTerminology · 15:00 — p95 10 msTerminology · 16:00 — p95 7 msTerminology · 17:00 — p95 7 msTerminology · 18:00 — p95 6 msTerminology · 19:00 — p95 6 msTerminology · 20:00 — p95 6 msTerminology · 21:00 — p95 6 msTerminology · 22:00 — p95 6 msTerminology · 23:00 — p95 6 msDe-identifyDe-identify · 00:00 — p95 8 msDe-identify · 01:00 — p95 9 msDe-identify · 02:00 — p95 8 msDe-identify · 03:00 — p95 9 msDe-identify · 04:00 — p95 8 msDe-identify · 05:00 — p95 10 msDe-identify · 06:00 — p95 9 msDe-identify · 07:00 — p95 9 msDe-identify · 08:00 — p95 9 msDe-identify · 09:00 — p95 10 msDe-identify · 10:00 — p95 12 msDe-identify · 11:00 — p95 15 msDe-identify · 12:00 — p95 18 msDe-identify · 13:00 — p95 19 msDe-identify · 14:00 — p95 22 msDe-identify · 15:00 — p95 26 msDe-identify · 16:00 — p95 23 msDe-identify · 17:00 — p95 21 msDe-identify · 18:00 — p95 18 msDe-identify · 19:00 — p95 15 msDe-identify · 20:00 — p95 13 msDe-identify · 21:00 — p95 10 msDe-identify · 22:00 — p95 9 msDe-identify · 23:00 — p95 9 msBulk exportBulk export · 00:00 — p95 32 msBulk export · 01:00 — p95 40 msBulk export · 02:00 — p95 43 msBulk export · 03:00 — p95 36 msBulk export · 04:00 — p95 34 msBulk export · 05:00 — p95 29 msBulk export · 06:00 — p95 23 msBulk export · 07:00 — p95 18 msBulk export · 08:00 — p95 18 msBulk export · 09:00 — p95 16 msBulk export · 10:00 — p95 16 msBulk export · 11:00 — p95 15 msBulk export · 12:00 — p95 14 msBulk export · 13:00 — p95 16 msBulk export · 14:00 — p95 14 msBulk export · 15:00 — p95 16 msBulk export · 16:00 — p95 14 msBulk export · 17:00 — p95 16 msBulk export · 18:00 — p95 16 msBulk export · 19:00 — p95 15 msBulk export · 20:00 — p95 17 msBulk export · 21:00 — p95 20 msBulk export · 22:00 — p95 24 msBulk export · 23:00 — p95 26 ms0006121823p95 6–45 ms

Enclave storage — used vs provisioned

200400-24MO-12MONOWPROVISIONED 320→420 TB349 TB · 83%
Platform operations console

What we deliver

  • Enclave architectures that inherit their controls from the hosting environment
  • Research enclaves for sensitive and limited datasets
  • Streaming and batch ingestion from legacy and modern sources
  • Clinical and claims data processing at population scale
  • NIST SP 800-171 aligned controls with automated evidence collection
  • Observability, monitoring, and documented incident response procedures

Health data engineering & interoperability

Data exchange is where health IT programs go to die. Hale MDS works natively in the standards — FHIR, HL7v2, X12, OMOP — and have connected systems that were never designed to talk. On top of that plumbing we design the intelligence: clinical decision support at the point of care, and AI-driven medical systems that survive contact with government oversight — documented, monitored, bias-tested, and explainable to a review board.

Hale MDS · Health-data interoperability

IFACE ENGINE · TERM SVCS · CDS

Feed volume

393k

messages/day · 5 live feeds

Code-map hit

99.8%

SNOMED↔ICD-10 · LOINC · RxNorm

End-to-end p95

1.1 s

feed arrival → CDS card

SOURCE FEEDSINTEGRATIONDECISION SUPPORTHL7V2 ADTadmits · transfersFHIR R4clinical resourcesX12 837 CLAIMSprof · institutionalLIS LAB FEEDresults · microIMMUNIZATION REGIIS · VXUINTERFACE ENGINEroute · transform · queueTERMINOLOGY SVCScode mappingCDS HUBrisk modelsEHR ALERTS · ORDERSQUALITY REGISTRIESANALYTICS ENCLAVE

Hover or tab through feeds, services, and links for protocol, daily volume, and p95 latency.

Clinical flowCode mappingMessage pulse
Health-data interoperability map

Hale MDS · Claims integrity

RISK MODEL · AUC 0.98

Claims embedding — anomaly surface

OUTPATIENT E/MIMAGINGPOST-ACUTE / DMEUMAP·1 →
Scored claimFlagged ≥ t (14)

ROC — drag the operating point

000.50.511FALSE POSITIVE RATETPRAUC 0.98t = 0.50TPR 83% · FPR 3%
Threshold0.50

Operating point at t = 0.50

Precision

69%

35 true of 51 flagged

Recall

83%

35 of 42 fraud caught

Flag rate

10.0%

51 of 512 claims held

FlaggedCleared
Fraud
35true pos
7false neg
Legit
16false pos
454true neg

Why this flag — feature contributions (top claim)

Billing velocity
+0.34
Prior denials
+0.27
Provider specialty mismatch
+0.22
Claim amount
+0.11
Patient distance
+0.07
Member tenure
−0.05
Network history
−0.12
Dx consistency
−0.19

base risk 0.08 → model score 0.83

Raises riskLowers risk
Claims risk-model evaluation console

What we deliver

  • Clinical decision support embedded in daily workflow
  • FHIR R4 servers, facades, and bulk-data pipelines
  • HL7v2, C-CDA, and X12 interface engineering
  • Terminology services: SNOMED CT, LOINC, RxNorm, ICD-10 mapping
  • Risk models with model cards, drift monitoring, and bias testing
  • Health IT modernization — incremental strangler-fig, not big-bang

How we work

Compliance is part of the study design, not a stage after it.

  • HIPAA

    Privacy by design

    De-identification, minimum-necessary access, and audit trails built into every data flow.

  • NIST SP 800-171

    Controls as code

    Security controls implemented as code with evidence collected automatically, so accreditation is not a retrofit.

  • Section 508

    Accessible by default

    WCAG 2.2 AA conformance designed in and tested with assistive technology.

See these capabilities applied to health missions.