Brian Wundheiler

Clinical and telemonitoring teams lacked reliable, consistent metrics to track patient engagement and adherence. Patient data arrived asynchronously from multiple devices, causing gaps, delays, and false non-submission alerts. Lastly, inconsistent definitions, such as active vs. inactive patients, led to unreliable dashboards and operational friction.

My approach started by clarifying operational decisions with clinical stakeholders and defining success metrics. KPI definitions were standardized—active vs. inactive patients, submitted today, days since last submission—including time zone and inclusion rules. Lightweight data contracts were defined and translated into a prioritized backlog with testable acceptance criteria. Finally, data quality checks and alert guardrails were added to prevent false non-submission alerts, and consistent BI reporting was enabled through a governed statistical dashboard.

Before, teams relied on inconsistent metrics and frequent false alerts. After, engagement KPIs were reliable, alerts decreased, and clinical teams prioritized patients with confidence using decision-ready data.

Fleet operators lacked reliable insight into real driving behavior. Existing telemetry data was available but not calibrated to public transport use cases, limiting its ability to support eco-driving analysis or trigger meaningful operational alerts.

I began by analyzing real-world bus operations and driver behavior patterns. Telemetry parameters were calibrated to match public transport dynamics, translating raw signals into eco-driving indicators. These indicators were then integrated into a monitoring platform with in-vehicle visual and audio alerts.

Before, fleet operators had limited visibility into driver behavior. After, calibrated telemetry enabled eco-driving analysis and real-time alerts, improving monitoring and supporting more efficient and safer public transport operations.

The AMIGA underground muon detector produced complex analog signals from scintillators and PMTs, but no analysis system existed to reliably identify signal patterns and count muons from raw detector data.

A dedicated data analysis system was developed to process raw detector signals, recognize characteristic patterns, and separate muon events from noise. Validation steps ensured accurate and reproducible muon counting as part of a doctoral research project.

Before, muon counting from underground detector signals was not feasible at scale. After, validated signal analysis enabled reliable muon identification, supporting scientific measurements and long-term data analysis.