Warfarin AI Dashboard – Clinical Decision Support

HEALTHCARE AND PHARMACOGENOMICS

Role: Business Analyst (Health Informatics)

Project Overview

Warfarin, a widely used anticoagulant, requires precise dosing due to narrow therapeutic windows and patient-specific risks. Incorrect dosing can lead to life-threatening complications such as stroke or bleeding events

The healthcare organisation aimed to enhance patient safety by developing an AI-assisted decision support dashboard that consolidates clinical, laboratory, and patient history data to help clinicians make more informed dosing decisions

The goal was to enhance dose accuracy, reduce review time, and support evidence-based decision-making by utilising a data-driven digital tool

Business Challenge

Clinicians faced significant limitations with the previous manual review process

Patient data was spread across multiple clinical systems

Dosage decisions required manual cross-checks of INR values, medication history, and clinical notes

High risk of inconsistent dosing decisions between providers

No real-time analytics or trend monitoring for high-risk patients

Limited ability to detect early signs of unsafe dosing patterns

This created clinical inefficiencies and increased patient risk.

Key Responsibilities

I served as the bridge between clinical teams, data scientists, and IT analysts, ensuring that requirements aligned with clinical workflows, safety standards, and technical feasibility

Stakeholder Engagement

Conducted interviews with clinicians, pharmacists, lab leads, and IT teams

Mapped roles and data touchpoints across the current Warfarin review process

Captured clinical pain points and decision-making criteria

Data & Requirements Analysis

Identified key data sources (INR history, dosage records, comorbidities, lab results)

Defined business rules and functional requirements for dashboard views

Documented clinical decision dependencies (red flags, thresholds, alerts)

Collaborated with data scientists to translate clinical logic into data models

Workflow & Solution Design

Mapped AS-IS Warfarin management pathway

Designed TO-BE workflow incorporating AI-supported recommendations

Created dashboard wireframes showing: Patient risk scoring, INR trend visualisations, Recommended dose ranges, Flags for unsafe conditions, Alert notifications

Defined user personas for different clinical roles (GP, nurse, pharmacist)

Prototyping & Visualisation

Built dashboard prototypes in Power BI

Created data visualisations for trend tracking, alerts, and risk stratification

Presented prototypes to clinical teams for iteration and validation

Testing & Validation

Developed UAT test scripts based on clinical scenarios

Coordinated user testing with nurses, pharmacists, and prescribers

Logged defects, usability concerns, and improvements for iteration

Delivery & Change Support

Documented operating procedures and dashboard guidelines

Produced training materials for clinical staff

Gathered feedback to support ongoing model improvement

Tools

Power BI

Excel

SQL (data extraction support)

Miro

Confluence

EHR/clinical system data feeds

Achievements & Impact

Reduced clinician review time by 40% through consolidated data views

Improved Warfarin dose safety through alert-based recommendations

Enabled the identification of high-risk patients earlier using trend analysis

Standardised the decision-making process across clinical roles

Increased confidence in dosing decisions through AI-supported insights

Delivered an analytics framework that can scale to other medications and chronic conditions

C Mary Agunroye 2025