Ontada transforms 150 million unstructured oncology documents with Azure OpenAI Service

Knowing each data point can make a life-changing difference for cancer patients. It is time-consuming, error-prone, and cost-prohibitive for Ontada to generate key insights using manual processes to analyze 150 million unstructured oncology documents.

Using Microsoft Azure AI Foundry and Azure OpenAI Service, Ontada implemented large language models to target nearly 100 critical oncology data elements across 39 cancer types, delivering access to quality data from unstructured oncology documents four times faster.

Ontada now accesses an estimated 70% of previously unanalyzed or unused information and has accelerated its life science product development, speeding up time to market from months to just one week.

Healthcare systems globally face a pressing data-related challenge: how to unlock the contents of volumes of unstructured, inaccessible healthcare data, estimated to represent 80% of health documentation, according to a 2019 Cornell University report. This unstructured data could be in the form of clinical notes, lab reports, medical images, sensor readings, genomics, and operational data. Given the enormity of processing this data, they are often left unanalyzed or analyzed on a small scale through manual abstraction.

Ontada sets out to improve real-world data for biopharma companies

Ontada, a McKesson business dedicated to oncology data and evidence, clinical education, and point-of-care technologies, set out to provide a more robust, real-world-focused data insights product for its biopharma customers. As a business within the McKesson Fortune 10 ecosystem of companies, Ontada has unmatched access to a large portfolio of oncology businesses and partners, each of whom provides research, insights, technologies, and services to help it advance oncological clinical education and point-of-care technologies on a national scale. McKesson considers Ontada a vital business driver.

“With cancer care becoming more targeted, providers, life sciences companies, and payers face a multitude of challenges and complexity in the development of new treatments and making them accessible to patients in need,” says Sagran Moodley, Senior Vice President and Chief Innovation Technology Officer at Ontada. “Unlike some data and insights companies that cover multiple therapeutic areas, Ontada specializes strictly in oncology, empowering healthcare providers to make informed decisions faster and work together towards more personalized patient care.”

Knowing that each data point can make a life-changing difference for its patients, Ontada began developing a comprehensive product capable of assisting biopharma companies in better understanding the patient’s journey, with a goal to accelerate drug development and treatment adoption at the time of launch. “We were collecting the right data, but it’s impossible for a human to make the correlations needed to gather meaningful insights at this scale,” says Amy O’Sullivan, PhD, Senior Vice President and Chief Research Officer at Ontada.

With each individual cancer patient as its inspiration, Ontada began looking for a scalable, automated, modern solution to unlock the secrets buried within more than 150 million unstructured oncology document components across its ecosystem continuum.

A modern platform helps catalyze cancer care breakthroughs

Ontada chose Microsoft to advance its AI-driven oncology research within an aggressive six-month journey.“Whereas traditional large language model solutions revealed scalability and data management barriers in processing that much data in such a short sprint, we knew the Azure data stack, security, and developer tools could help Ontada uncover and deliver the life-saving biopharma breakthroughs we knew were possible,” says Moodley.

We knew the Azure data stack, security, and developer tools could help Ontada uncover and deliver the life-saving biopharma breakthroughs we knew were possible.

Sagran Moodley, SVP and Chief Innovation Technology Officer, Ontada

In collaboration with Microsoft, Ontada began developing its novel data platform solution called ON.Genuity. Using Microsoft Azure AI Foundry and the Azure OpenAI Service Batch API to accelerate innovation, Ontada implemented large language models to target nearly 100 critical oncology data elements across 39 cancer types, significantly increasing its ability to rapidly extract important data and clinical details.

“The Azure OpenAI Service Batch API enables us to apply an elevated rigor and unparalleled computing power, delivering exceptional data quality, accuracy, and deep insights into the longitudinal patient journey,” says Wanmei Ou, PhD, Vice President of Data Analytics and Artificial Intelligence at Ontada. “With this transformative tool, we are building a comprehensive solution that redefines innovation in healthcare, powering new perspectives and possibilities for life science R&D and patient care.”

The Azure OpenAI Service Batch API enables us to apply an elevated rigor and unparalleled computing power, delivering exceptional data quality, accuracy, and deep insights into the longitudinal patient journey.

Wanmei Ou, PhD, VP of Data Analytics and Artificial Intelligence, Ontada

Ontada unearths unparalleled clinical data

To build its ON.Genuity platform, Ontada combined the innovation of generative AI’s sophisticated common data model with insights embedded in its proprietary electronic health record system called iKnowMed.

Because electronic health records are structured, and medical notes or scan reports are unstructured, Ontada’s ON.Genuity platform first pulls the structured data into the data lake, followed by the unstructured data, then integrates it all into a consolidated 360-degree view.

“We use Azure Databricks for our data pipeline to process both types of data, then Azure Document Intelligence to extract text from PDFs and other documents,” explains Ou. “This data is then processed by Azure OpenAI to extract specific entities like tumor histology, cancer staging, and other important patient details.”

“We were able to seamlessly combine the capabilities of Azure OpenAI with our extensive physician network and create a system that excels at patient trial matching, AI-driven insights, and care gap identification,” says O’Sullivan. “We also embraced generative AI to better predict outcomes with what/if analysis after extracting the unstructured document processing, including all the variables within."

We were able to seamlessly combine the capabilities of Azure OpenAI with our extensive physician network and create a system that excels at patient trial matching, AI-driven insights, and care gap identification.

Amy O’Sullivan, PhD, SVP and Chief Research Officer, Ontada

A single platform delivers speed and compliance

The ON.Genuity platform approach made it possible for Ontada to provision the entire platform and receive a data analysis report in under 45 minutes. “What could have taken weeks was done in half an hour, an efficiency that was crucial given our tight timeline and large data volume,” says Moodley.

"By using Microsoft Azure OpenAI Service Batch API, we processed 150 million documents in three weeks, reducing our processing time by 75%,” explains Ou. “This enables us to keep up with fast-moving advancements in science to find the most appropriate treatment solutions for improved patient outcomes. Our unique position in the community-based oncology space allows us to create a learning health system."

The Ontada model adheres to rigorous FDA guidelines, measuring against highly curated and manually curated data to ensure its model exceeds the sensitivity and specificity requirements set by the FDA.

More about this diagram

The future of cancer prevention looks brighter

For Ontada, Azure AI opens up the potential to exponentially increase its work as an evidence-based medicine practice, personalizing care and drug development for patients everywhere. Using Microsoft Azure AI Foundry, Ontada has already accelerated its life science product development, speeding up time to market from months to just one week.

“Our strategic collaboration with Microsoft enables us to extend the boundaries of AI-driven oncology research by generating meaningful and actionable data at unprecedented scale and speed,” says Moodley.

“The outcomes we’re seeing are absolutely incredible,” concludes O’Sullivan. “We’re closer to the patient journey than ever before, and the progression we’re seeing with AI adoption gives hope to the millions  fighting cancer globally.”

Discover more about Ontada on LinkedIn, and X/Twitter .