Google LLC’s cloud division today announced several new generative artificial intelligence solutions aimed at healthcare and life sciences aimed at enhancing data interoperability, search and improve patient outcomes.

The announcement, which came during the Healthcare Information and Management Systems Society Inc. 2024 annual conference, Google Cloud unveiled new features with Vertex AI search, a healthcare data platform and new capabilities for the company’s MedLM generative AI foundation models.

During 2022, the administrative costs burdening healthcare rose by 30%, according to the CAQH Index report, and the World Health Organization reported that there was a shortage of more than 13 million nurses globally in 2020. Even post-pandemic, healthcare organizations are still struggling to keep up with patients and data, so it’s a good time for them to up their game when it comes to data and outcomes.

Google Cloud said its Vertex AI Search for Healthcare, available today, will help developers build better systems for health systems workers to deal with administrative issues and vast numbers of clinical records. It will allow developers to connect search and generative AI conversational question-answer capabilities to a broad spectrum of data across healthcare systems, including using MedLM, Healthcare Data Engine and Cloud Healthcare application programming interfaces for access to data.

Among the key features, Vertex AI Search will include medically tuned search capabilities that can understand electronic health record data, scanned documents and other clinical data input by nurses, doctors and other clinicians. This will mean that doctors and other healthcare workers won’t need to shuffle through numerous documents and notes themselves just to find a single page or application to discover what they’re looking for, they can just type in what they need and have it found quickly.

With the power of generative AI at their fingertips, developers can create chatbots that can ingest all of the data above and allow question-answering capabilities and conversational services provided. This means that doctors and nurses can “talk” to their records with chatbots that can understand medical jargon and answer questions such as “Does this patient have a history of cancer?” The application can then answer in a way that understands medical terminology and also provides context for the medical notes in the scanned documents.

“Not all generative AI is created equal, and in healthcare, the stakes are particularly high,” said Aashima Gupta, global director for healthcare strategy and solutions at Google Cloud. “Healthcare organizations require enterprise-grade gen AI solutions, grounded in real data. Vertex AI Search for Healthcare is already making a difference for healthcare organizations by helping ensure clinicians have the right information and insights at the right time.”

Healthcare Data Engine expands globally

All of this is interoperable with Google Cloud’s Healthcare Data Engine, the way that organizations best take advantage of generative AI by providing access to high-quality data. Google Cloud today announced a new consumption pricing for the service, expanded availability globally and other features for customers.

Using the Healthcare Data Engine, healthcare organizations can upload patient records to the cloud at scale. This allows search, analytics and generative AI capabilities to get access to them in real-time and reduces the time needed to ingest and access them. The service is also compliant with regulations such as the Health Insurance Portability and Accountability Act, or HIPAA.

Users can now deploy HDE on a consumption-based pricing model, Google said and pay-as-you-go, using it as a managed service, which will open it up to more healthcare organizations to take advantage of its capabilities. Also, Google said that the data engine will be expanding beyond North America and rolling out globally to most cloud regions in Europe and the Asia Pacific, providing customers with global access.

Developers have been given access to a new low-code graphical tool that will help them transform their existing data into Fast Healthcare Interoperability Resources, an interoperable standard for exchanging healthcare data. By transforming it into this format it can be used to interact with more systems and help form more complete pictures of what’s happening in patient’s lives.

MedLM and greater functionality

MedLM, Google Cloud’s AI foundation models fine-tuned for the healthcare industry, will receive two new features that customers can test including chest x-ray classification and a task-specific API for condition summarization.

MedLM is a family of foundation models that Google introduced for healthcare that are available to Google Cloud customers through the Vertex AI platform. Currently, there are two models, built on Med-PaLM 2, that are designed for approaching different tasks such as conversational needs or searching for the correct medication.

The first new capability is MedLM for Chest X-ray, which will help with the classification of chest X-rays for technicians and doctors in screening and diagnostic use cases. It is a domain-specific model, that exists as an API that converts images of X-rays into embeddings, which are then used by the generative AI model. These embeddings can then be used by app developers along with classification for training and fine-tuning to assist in identifying conditions.

Google said the second new capability is a task-specific API called “Condition Summary,” which will provide a chronological list of patient conditions with AI-generated briefs for each condition. The list will supply citations including original text, so that clinicians have transparent ground truth that they can refer back to patient records and healthcare documents.

Google added that the company plans to bring even further functionality to the MedLM family of models in the coming months.

Image: Shutterstock/Everything Possible