Challenged to realize the full potential of its data, a leading U.S.-based healthcare company partnered with CGI to explore and build solutions that leverage AI and GenAI to drive better outcomes and lower costs for patients, providers and health plans.
The challenge
A U.S.-based healthcare company that provides medical benefits management faced significant challenges in meeting its mission to enable evidence-based care decisions. Hindered by the stability and quality of its data, hundreds of data points and manual processes, the company needed to improve its operational efficiency and enable faster and more informed decision-making.
To tackle this challenge, the company turned to CGI for guidance on maximizing its data's full potential. The partnership enabled the team to focus on increasing data availability and providing the tools to build and deploy machine learning models powered by GenAI for multiple use cases to improve health services.
It starts with data
CGI began by migrating the healthcare company’s data lake into the Databricks cloud-based platform on Azure. By refactoring the data and moving it to the cloud, the overall quality and stability of the data pipelines immediately increased.
CGI’s AI LaunchPad—accelerating AI value
Next, to define its AI strategy, CGI supported the company with its proprietary framework for end-to-end AI capabilities to mitigate risk and leverage GenAI responsibly. The framework includes four imperatives for action, powered by CGI’s AI LaunchPad, which accelerates AI projects, resulting in faster, more impactful and responsible AI deployments to drive business value.
Imperatives for action:
- Envision: Define your AI vision
- Explore: Identify high-impact, ROI-led use cases
- Engineer: Build a scalable and adaptable AI foundation
- Expand: Accelerate value and operate responsibly
The journey began with the envision phase, during which the teams worked together to identify, assess and prioritize AI use cases. They also explored challenges and opportunities to shape the company’s AI vision.
In the explore phase, subject matter experts and data scientists collaborated to evaluate and develop proof-of-concept solutions—such as predicting oncology regimen costs—demonstrating how AI can process and synthesize vast amounts of data.
As the company matured on its AI implementation journey, it moved into the engineer and expand phases, focusing on improving patient care by connecting patients, providers and health plans. Leveraging GenAI, the team identified and developed impactful use cases to drive optimized healthcare outcomes. To streamline processes and optimize cost savings, the CGI team consistently made recommendations that the technology organization reviewed and adopted across multiple workflows. The close collaboration between CGI and the engineers was key to successful implementation across use cases.
CGI enabled the healthcare company to explore opportunities to leverage AI and GenAI technology, building models and solutions that evaluate the accuracy of AI responses over time—resulting in critical business insights.
Accelerating business value with three use cases
AI-powered oncology cost forecasting and regimen reclassification
| Challenge: | Solution: | Outcomes: |
|---|---|---|
|
To enhance cancer patient care, the company needed a better way to predict oncology regimen costs and identify mislabeled regimens in patient claim authorizations. Accurate predictions would improve budgeting for approved expenses, streamline workflows and optimize costs, ultimately increasing efficiency and saving time. |
CGI helped develop and implement an AI-powered cost forecasting model trained on historical oncology cases. By recognizing hidden patterns, it accurately projected total costs for current cases. Additionally, the team delivered a cancer regimen reclassification model that analyzed past authorizations to detect and correct mislabeled regimens, ensuring alignment with standard classifications. |
With these AI-driven models, the company can now more effectively budget and plan. Their enhanced ability to reduce foreseeable oncology costs while ensuring accurate clinical data classification results in a regimen classification accuracy of 85%. |
Clinical GenAI chatbot frees clinicians from manual processes and saves US$2.5 million annually
| Challenge: | Solution: | Outcomes: |
|---|---|---|
| The healthcare company needed to streamline its burdensome claims review process, which required nurses and doctors to search through a large number of faxed documents manually. This tedious and time-consuming process prevented clinicians from focusing on delivering patient care. | CGI developed an AI-powered chatbot to assist clinicians in extracting key information from faxes, enabling quicker case decisions. Instead of manually searching through documents, clinicians could ask domain-specific questions like, “What medications are listed?” or “What are the patient’s lab results?” and receive answers instantly, eliminating time searching and scrolling. | The interactive AI chatbot reduces clinicians' manual effort, delivering concise and accurate information from large documents for faster, data-driven decision-making. This led to higher accuracy in claim approvals or denials, fewer customer service calls and an annual savings of US$2.5 million. |
Predicting physician calls with live call models
| Challenge: | Solution: | Outcomes: |
|---|---|---|
| Every time a physician called the healthcare benefits management company regarding a patient case, the associated support costs increased. With rising expenses per case, the company aimed to reduce unnecessary calls to its call center. | CGI developed AI-powered predictive models to assess the likelihood of physicians calling during various health plan adjudication steps. The company identified the key reasons behind 80% of calls using live call prediction models and AI-powered analytics dashboards. This allowed them to proactively provide information, reducing the need for physician-initiated inquiries. | The predictive models enabled significant time and cost savings per case while optimizing call center operations. By predicting and eliminating the need for multiple calls, physicians can focus on higher-value tasks that improve patient care. |
Delivering powerful business outcomes
By leveraging AI and GenAI-powered solutions, the company can now rapidly and accurately analyze large, complex datasets, enabling data-driven decision-making in key areas such as cost prediction, regimen recommendations and claims review. These advanced AI solutions have delivered significant benefits, including:
- Optimized costs: Predictive analytics to forecast and reduce expenses.
- Increased efficiency: Automated workflows and AI-driven insights to streamline operations.
- Enhanced decision-making: Greater data accuracy and AI-supported decision processes across the organization.
- Improved services: Comprehensive models incorporating clinical, social and geographic data to drive evidence-based outcomes.
Working with a trusted partner for continued success
With CGI as its trusted partner, the outlook for improving medical benefits management through innovative, value-added service capabilities is promising. The healthcare company has continued developing additional use cases—powered by robust GenAI solutions—enabling key operational insights that drive unprecedented business outcomes.
In addition to empowering medical benefits management, CGI’s AI LaunchPad helps organizations rapidly identify, design, validate and deploy AI solutions. Other industries—fleet management, wealth management, retirement and life insurance, utilities, home building and telecommunications— successfully leverage AI LaunchPad to enable AI-powered business models.
In addition to empowering medical benefits management, CGI’s AI LaunchPad helps organizations rapidly identify, design, validate and deploy AI solutions. Other industries—fleet management, wealth management, retirement and life insurance, utilities, home building and telecommunications— successfully leverage AI LaunchPad to enable AI-powered business models.
