Clinical Trial Safety & Risk Analytics (IRMS): Overhauling Enterprise Surveillance for a Fortune 50 BioPharma Giant
How we transformed rigid, error-prone TIBCO Spotfire BI dashboards into a unified clinical surveillance platform for over 100+ global medical coordinators, resolving systemic production defects and cutting clinical trial setup & adverse review cycles by ~40%.
Role
Lead UX / Systems Architect
Company
HashedIn by Deloitte
Primary Users
Medical Reviewers & Epidemiologists
Measured Impact
~40% Faster Review
"Great work Ujjwal!"
Direct commendation from Global BioPharma Client Leadership upon receiving the redesigned Clinical Risk Adverse Report 9.
The Context: High-Stakes Patient Safety
During Phase II and Phase III global clinical drug trials, medical directors and safety coordinators must continuously monitor thousands of trial sites and patient cohorts. A single missed adverse signal or protocol deviation can lead to patient harm, regulatory holds from the FDA/EMA, and multi-million-dollar trial cancellations.
The Core Paradox:
While millions of data points were successfully stored in underlying databases, clinicians were forced to analyze them through out-of-the-box TIBCO Spotfire BI dashboards that were designed for generic financial spreadsheets, not high-velocity medical risk triage.
1. The "Floating Filter" Bug
Spotfire’s dynamic filter panels floated over analytical canvases, physically covering chart axes, legends, and patient ID columns on every single report.
2. Unreadable Risk Plots
Standard box plots failed to communicate trending Risk Priority Numbers (RPN). Clinicians couldn't tell if a high score was an isolated anomaly or an accelerating risk trend.
3. Lost Breadcrumb Context
When clinicians drilled down into a specific adverse patient record, the system wiped their cohort filters, forcing them to re-apply 8-10 filter criteria from scratch.
Discovery & Mental Model Mapping
Through stakeholder interviews with global clinical trial directors and biopharma safety physicians, we mapped the clinical review mental model from an unstructured search to a rigorous 3-tier triage funnel:
"Which of my 14 ongoing oncology studies has an unusual spike in adverse events this week?"
"Is the spike concentrated in European clinics or specific age cohorts receiving 50mg dosage?"
"What was Patient #8492's vitals trajectory 48 hours prior to the reported toxicity?"
Architectural Solutions & Craft
1. Dockable Filter Drawer & Custom Navigation Shell
System-Wide Defect FixRather than accepting Spotfire's native DOM limitations, Ujjwal designed a custom outer layout shell featuring a persistent, collapsible filter drawer with saved search presets.
• Severity: Grade 3+ [x]
• Cohort: 40-65y [x]
• Region: Global (All)
[Save Bookmark Preset]
Longitudinal RPN Trendlines with Confidence Thresholds (No floating popups blocking visual space).
Outcome: Eliminated 100% of reported screen occlusion tickets across all production IRMS reports, establishing the UX standard for subsequent releases.
2. Longitudinal Trending RPN Analytics (Report 9 Redesign)
Client CommendationReplaced static box-plots with a dual-axis time-series visualization. Clinicians could now evaluate Risk Priority Numbers (RPN) over consecutive patient visits (Baseline ➔ Week 4 ➔ Week 12 ➔ Week 24) overlaid against institutional safety thresholds.
- Delta Badges (ΔRPN): Immediately highlight cohorts whose risk velocity exceeds +15% week-over-week.
- Outlier Inspection Drawer: Clicking any high-risk data point slides out patient demographics without leaving the comparative trend graph.
- 4-Color Stacked Bar Standard: Standardized Report 6 and Report 26 to strict color-blind compliant regulatory palettes (FDA/EMA submission ready).
3. Cross-Firm UX Data Dictionaries & Trackers
Process InnovationTo prevent engineering misinterpretation across offshore teams, Ujjwal pioneered a standardized UX Data Dictionary:
| UI Element | Clinical Schema Field | Null / Loading State Behavior | Validation Rule |
|---|---|---|---|
| RPN Delta Chip | `delta_rpn_30d` | Display neutral badge ("N/A") | Trigger amber alert if > 15.0 |
| Cohort Size Badge | `active_patient_count` | Display skeleton shimmer | Require minimum N=5 for anonymization |
Coupled with a weekly Prototype Review Tracker, sprint revisions and QA bugs dropped to zero during production handoffs.
4. Multi-Stream Leadership & Mentorship
LeadershipSimultaneously directed 3 parallel design streams (Reporting Redesigns, Release 3.1 & 4.0 Core Architecture, and the ASRP Review Tool) while coaching junior designer Heli to run client workshops independently.
Measurable Business & Clinical ROI
~40%
Faster Clinical Trial Review
Accelerated weekly safety sign-offs across global cohorts
100%
Production Reports Fixed
Eliminated the floating filter panel bug across all views
0
Handoff Blockers
Zero schema mismatches via UX Data Dictionaries
Retrospective & Design Leadership Reflections
What I learned: In enterprise healthcare platforms, design is not cosmetic styling—it is a safety buffer. When data hierarchy is ambiguous, users miss critical signals. By establishing data dictionaries and listening closely to medical reviewers' daily friction, we aligned complex engineering with high-stakes human responsibility.
What I would do differently: If starting over, I would introduce code-level component tokens directly into the Spotfire template layer earlier, which would have saved an additional 2-3 weeks of front-end styling alignment during Release 1.