Project

Support Optics – Finance, BI & Healthcare Analytics

Multi-Industry Analytics · HIPAA-Compliant Healthcare Insights · Predictive Modelling · BI Dashboards

Support Optics January 2013 – January 2016 Data Analyst

Executive Summary

Delivered actionable insights for clients across finance, business intelligence, and healthcare through in-depth data analysis and research. In healthcare, analyzed large datasets identifying patterns and quality improvement opportunities while maintaining full HIPAA compliance; developed predictive models forecasting healthcare outcomes including HEDIS measure tracking and care gap analysis. In finance, designed data collection methodologies improving accuracy and reliability for investment strategy and resource allocation decisions. Built interactive dashboards using Power BI, Tableau, and Excel enabling clients to track KPIs and improve operational efficiency. Performed statistical analysis including regression, hypothesis testing, and clustering to extract actionable insights.

Where My Data Career Really Began

Support Optics was where I learned that data without context is just noise. My first assignment was analysing hospital readmission rates for a healthcare client. I built a beautiful SQL query that identified patients with high readmission risk—but when I presented the results, the clinical team told me my "high-risk" cohort was actually patients receiving planned follow-up procedures. The data was technically correct but clinically meaningless. That humbling experience taught me the most important lesson of my career: always understand the domain before you trust the numbers. I spent the next two weeks shadowing clinical operations staff, learning the difference between planned readmissions, emergency readmissions, and observation stays. The model I built after that domain immersion actually saved the client money—because it targeted the right patients.

Healthcare Analytics

HIPAA-compliant analysis of large healthcare datasets; predictive models for patient outcomes; HEDIS measure tracking; care gap identification; readmission reduction analytics; HL7 data integration.

Finance & BI

Financial trend analysis and data collection methodologies; interactive Power BI, Tableau, and Excel dashboards; KPI tracking; resource allocation optimization; investment strategy support.

Statistical Analysis

Regression analysis, hypothesis testing, and clustering; data validation and quality assurance; predictive modelling for healthcare and financial outcomes across multiple client engagements.

Business Challenges

The HIPAA Lesson That Changed My Approach to Data Security

Working with healthcare data taught me that compliance isn't a checkbox—it's a mindset. Early on, I needed to share a readmission analysis with a clinical operations team. I de-identified the data by removing patient names but left dates of birth and zip codes. A colleague caught it during review: the combination of DOB + zip code + admission date could potentially re-identify patients in small rural hospitals. That near-miss rewired how I think about every dataset. Now I apply a "minimum necessary" principle everywhere—even in non-healthcare contexts. At Signet years later, when handling M&A financial data, I instinctively applied row-level security and data masking on deal-sensitive fields. Healthcare compliance training made me a better data professional across every industry.

Healthcare Quality Analytics

HEDIS measure tracking, care gap analysis, readmission risk scoring, and CMS Star Rating dashboards for healthcare clients.

πŸ₯
HEDIS Measures Tracked
42
Across 5 domains
πŸ“ˆ
Care Gap Closure Rate
73.8%
+18.2% improvement
⭐
CMS Star Rating
4.2
Up from 3.5
πŸ”’
HIPAA Compliance
100%
Zero incidents

HEDIS Quality Measure Performance

Breast Cancer Screening
84% Compliant
Diabetes HbA1c Control
78% Controlled
Colorectal Screening
71% Compliant
Blood Pressure Control
67% Controlled
Medication Adherence
58% Adherent
Above Benchmark (>65%) Below Benchmark

The Medication Adherence Discovery

That 58% medication adherence rate? It was the starting point for one of my most impactful analyses. The clinical team assumed non-adherence was a patient behaviour problem. My clustering analysis told a different story: 62% of non-adherent patients had the same pharmacy benefit plan that required mail-order refills with a 14-day processing window. Patients were running out of medication during the gap. The "non-compliance" wasn't a patient problem—it was a system design problem. The client renegotiated their pharmacy benefit contract to allow retail bridge prescriptions, and adherence jumped to 74% in the next measurement period. I learned that before you label a data pattern as "behaviour," investigate the systems that create the pattern.

Solution Architecture

  1. Healthcare Data Integration (HIPAA): Built HL7-compliant data pipelines integrating clinical, claims, and pharmacy data from Epic, Cerner, and Allscripts EHR systems; implemented PHI de-identification protocols and access controls ensuring full HIPAA compliance
  2. HEDIS & CMS Star Tracking: Developed automated HEDIS measure calculation engine covering 42 measures across preventive care, chronic disease management, and patient experience domains; mapped clinical data to CMS Star Rating methodology
  3. Predictive Healthcare Models: Built predictive models using regression and clustering techniques to forecast patient outcomes; modelled readmission risk, care gap probability, and chronic disease progression for proactive intervention
  4. Financial Analytics Platform: Designed data collection methodologies and validation rules for finance clients; built dashboards tracking investment performance, resource allocation efficiency, and financial trends
  5. BI Dashboard Suite: Created interactive dashboards in Power BI, Tableau, and Excel for clients across all three verticals; standardized KPI definitions and implemented data quality assurance ensuring accuracy and consistency
  6. Statistical Analysis Practice: Performed regression analysis for outcome prediction, hypothesis testing for intervention effectiveness, and clustering for patient/customer segmentation; delivered actionable recommendations grounded in statistical rigour

Why I Still Think in Star Schemas

Support Optics was where I first learned dimensional modelling—not from a textbook, but from necessity. Our healthcare client needed a report that joined patient demographics, clinical encounters, lab results, prescriptions, and HEDIS gap status. My first attempt was a 400-line SQL query with 12 JOINs that took 45 minutes to run. A senior colleague looked at it and said: "You need a star schema." I redesigned the data model with a patient dimension, encounter fact table, and bridge tables for the many-to-many relationships. Same report, 8 seconds. That experience fundamentally changed how I think about data. Even today at Signet and AT&T, I start every project by asking: "What are the facts? What are the dimensions?" The dimensional modelling foundation I built at Support Optics has been the backbone of every BI platform I've designed since.

Sample SQL — Care Gap Identification

SQL — HEDIS Care Gap Detection Query

-- Identify patients with open care gaps for preventive screenings
WITH eligible_patients AS (
    SELECT
        p.patient_id,
        p.age,
        p.gender,
        p.pcp_provider_id,
        p.risk_score
    FROM analytics.dim_patients p
    WHERE p.is_active = TRUE
      AND p.age BETWEEN 50 AND 75
      AND p.enrollment_status = 'ACTIVE'
),
completed_screenings AS (
    SELECT DISTINCT
        e.patient_id,
        e.procedure_code,
        e.encounter_date
    FROM analytics.fct_encounters e
    WHERE e.procedure_code IN ('77067', '77066', '45378', '45380')  -- Mammogram, Colonoscopy
      AND e.encounter_date >= DATEADD('year', -2, CURRENT_DATE)
      AND e.claim_status = 'PAID'
)
SELECT
    ep.patient_id,
    ep.pcp_provider_id,
    ep.risk_score,
    CASE WHEN cs_mammo.patient_id IS NULL AND ep.gender = 'F'
         THEN 'OPEN' ELSE 'CLOSED' END AS breast_cancer_screening_gap,
    CASE WHEN cs_colon.patient_id IS NULL
         THEN 'OPEN' ELSE 'CLOSED' END AS colorectal_screening_gap
FROM eligible_patients ep
LEFT JOIN completed_screenings cs_mammo
    ON ep.patient_id = cs_mammo.patient_id
    AND cs_mammo.procedure_code IN ('77067', '77066')
LEFT JOIN completed_screenings cs_colon
    ON ep.patient_id = cs_colon.patient_id
    AND cs_colon.procedure_code IN ('45378', '45380')
WHERE cs_mammo.patient_id IS NULL
   OR cs_colon.patient_id IS NULL
ORDER BY ep.risk_score DESC;

The Query That Taught Me About Healthcare Data Quality

That care gap query looks clean now, but my first version returned 3x more "open gaps" than actually existed. The problem? I was only checking clinical encounter data and missing pharmacy claims (some screenings are captured differently), supplemental data submissions, and patients who received care outside the network. Healthcare data is never in one place. I learned to build multi-source gap closure logic that checks encounters, claims, pharmacy, and supplemental data feeds. "Trust but verify" became my mantra—and that principle now applies to every data pipeline I build, healthcare or not. The financial reconciliation framework I later built at Signet used the same multi-source verification approach.

Results

Healthcare Quality Improvement

HEDIS measure compliance improved across all tracked domains; care gap closure rate increased by 18.2%; CMS Star Rating improved from 3.5 to 4.2 for primary client; predictive readmission models enabled proactive patient outreach.

Financial Analytics Impact

Improved data collection methodologies increased reporting accuracy and reliability; financial trend dashboards supported resource allocation optimization and investment strategy decisions for finance clients.

BI & Statistical Insights

Interactive Power BI, Tableau, and Excel dashboards enabled clients across finance, BI, and healthcare to track KPIs and improve operational efficiency; statistical analyses (regression, clustering, hypothesis testing) drove data-backed business strategy improvements.

The Foundation That Built Everything After

Looking back, Support Optics was the most formative experience of my career. Not because of the technology—we were using SQL Server and SSIS, tools I've long since moved past. But because of the lessons: HIPAA taught me data governance before "data governance" was a buzzword. Healthcare complexity taught me to always understand the domain. Finance clients taught me to quantify impact in business terms. The multi-industry exposure forced me to be adaptable—switching between healthcare compliance, financial analysis, and operational BI within the same week. Every role I've held since—Syilum's product analytics, AT&T's enterprise BI, Signet's M&A intelligence—has been built on skills I first developed at Support Optics.

Challenges & Lessons

Tech Stack

Data & Analytics: SQL (SQL Server, SSIS), Python, Statistical Modelling (Regression, Hypothesis Testing, Clustering), Predictive Analytics
BI & Visualization: Power BI, Tableau, Advanced Excel (Pivot Tables, VBA), KPI Dashboards
Healthcare: HIPAA Compliance, HEDIS Measure Tracking, CMS Star Ratings, HL7 Integration, Epic/Cerner/Allscripts EHR Data
Domains: Healthcare Analytics, Finance & Investment Analysis, Business Intelligence, Data Quality Assurance