Executive Summary

A leading healthcare organization sought to improve cardiometabolic therapy initiation by identifying patients most likely to begin treatment and providing clinicians with real-time, AI-generated insights. Manual review of patient records and unstructured clinical notes had been time-consuming and inefficient, delaying therapy decisions.

By leveraging Quantiphi’s AI-driven solution built on Google Cloud’s Vertex AI, Gemini, and BigQuery, the organization developed an end-to-end predictive model that identified high-probability patients for therapy initiation while automatically summarizing clinical notes with LLM-based intelligence. This enabled faster, data-backed physician decisions and improved patient engagement across the care continuum.

About the Client

A healthcare organization operating a digital health coordination unit dedicated to improving patient triaging, therapy initiation, and clinician efficiency using AI and predictive analytics.

Industry: Healthcare
Size: Large enterprise healthcare provider (over 10,000 employees)
Country: USA

Problem Statement

The organization faced growing challenges in identifying eligible patients likely to initiate cardiometabolic therapy due to the manual and fragmented nature of data review. Physicians spent significant time parsing through unstructured clinical notes, delaying interventions and treatment decisions. Without predictive analytics or automated summarization tools, the triaging process lacked real-time insights necessary for proactive patient management. The organization needed an AI-driven, integrated solution to deliver predictive insights, streamline triaging, and enhance provider efficiency.

Challenges

Difficulty identifying eligible patients most likely to begin cardiometabolic therapy.
Time-intensive manual review of clinical notes limited decision-making efficiency.
Absence of real-time predictive insights integrated into provider workflows.
Lack of automated, structured reporting for therapy prioritization.

The Solution

Quantiphi partnered with the organization to design a cloud-based, AI-driven propensity modelling solution built on Google Cloud technologies. The solution integrated structured and unstructured patient data into a unified intelligence layer, generating real-time therapy initiation predictions and automated clinical note summaries to support informed physician decisions.

Key components included:

Predictive Propensity Modelling: Identified patients most likely to initiate cardiometabolic therapy using Vertex AI models trained on historical health and behavioral data.
LLM-Powered Clinical Summarization: Deployed Gemini (Google’s LLM) to interpret and summarize unstructured clinical notes for faster provider review.
Looker Dashboard Integration: Developed interactive dashboards visualizing patient demographics, risk profiles, and summarized insights for clinical decision support.
Data Infrastructure: Leveraged BigQuery and Cloud Storage for scalable, secure, and compliant data handling.

This end-to-end pipeline empowered the client’s digital health team to triage patients efficiently, make real-time intervention decisions, and reduce administrative burden on healthcare providers.

Impacts

Enabled real-time, data-driven triaging for cardiometabolic patients.
Automated clinical summarization, reducing manual note review time.
Enhanced provider efficiency and patient engagement through predictive insights.
Identified 500+ patients likely to initiate therapy using AI propensity scores.
Established a scalable, cloud-native foundation for future digital health applications.

Technologies Used

Google Cloud Vertex AI

Gemini LLM

Looker

BigQuery

Google Cloud Storage

Results

Increased identification of patients likely to begin cardiometabolic therapy.
Improved provider efficiency through AI-driven, automated summarization workflows.
Delivered real-time, actionable insights integrated into clinical dashboards.
Established a reproducible framework for patient triaging and therapy prediction.