AI for Healthcare

AI is an increasingly critical component of our healthcare ecosystem, and its potential applications are vast. Right now, AI is already being used to detect diseases earlier and more accurately. AI is also streamlining drug research and discovery processes in ways that could significantly reduce drug costs and the time it takes to bring new drugs to market. AI can help track the outbreak and spread of diseases more efficiently and effectively, allowing public health experts to engage in real-time modeling for managing and stopping epidemics and pandemics.

Detecting Alzheimer’s Earlier

RetiSpec

Current techniques for identifying Alzheimer’s disease are expensive and typically involve a spinal tap or a PET scan.  RetiSpec, a medical imaging company, developed an AI system that can analyze eye scans and detect signs of Alzheimer’s up to 20 years before symptoms appear.  RetiSpec’s AI reads scans from a camera that can be attached to machines already available in most optometrists’ offices.  The camera measures a wider range of the spectrum than the human eye can see, which allows the AI to detect unique optical signatures that correspond with the presence of abnormal levels of amyloid in the brain, which is associated with Alzheimer’s disease.  The model, which delivers results instantaneously with 80% accuracy, could make early detection of Alzheimer’s much more affordable and accessible, leading to earlier interventions and treatments for many individuals, particularly those who cannot afford expensive out-of-pocket diagnostic exams.

 

Improving Early Detection of Pancreatic Cancer

MIT’s PRISM

Pancreatic cancer is a difficult disease to detect.  Patients rarely experience symptoms in the early stages, meaning that most cases are diagnosed at an advanced stage, making it much harder to cure.

Researchers from MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) developed an AI system that predicts a patient’s likelihood of developing pancreatic ductal adenocarcinoma (PDAC), the most common form of pancreatic cancer.  This new AI-based risk prediction system could help catch deadly pancreatic cancer cases earlier.

The new AI system, PRISM, uses artificial neural networks to spot patterns in medical data, including patients’ ages, medical history, and lab results, to determine individual risk scores for developing pancreatic cancer.  The PRISM model was trained on anonymized data from six million electronic health records, 35,387 of which were PDAC cases, from 55 healthcare organizations in the U.S.  The AI system outperformed current diagnostic standards, identifying 35% of those who developed pancreatic cancer as high-risk six to 18 months before their diagnosis.  The current standard screening criteria catch around only 10% of cases.

 

Meeting Pandemic Demands

Duke University and Microsoft AI for Health

During the height of the COVID-19 pandemic, Duke University partnered with Microsoft AI for Health to quickly develop a new ventilator splitter and resistor system to help COVID-19 patients, as a growing number of hospitals continued to face a shortage of ventilators. In order for the system to work, each device had to be customized to each patient. Patients sharing the splitting system had to be approximately the same weight and have similar lung compliance. This required scientists to run hundreds of millions of air simulations using nearly a million compute hours. Using Microsoft Azure AI, the team was able to complete 800,000 compute hours within a short amount of time, enabling them to achieve an FDA emergency use authorization for the device within a week’s time.

 

Advancing Identification of Diseases

Deepmind

Deepmind’s AlphaMissense AI model has significantly advanced our understanding of one of the greatest challenges in human genetics – how we identify the root cause of diseases. AlphaMissense has the ability to make accurate predictions about “missense” mutations or “misspellings,” in human DNA code. These mutations are often harmless, though they can disrupt how proteins work and, in some cases, they can lead to diseases such as cystic fibrosis, sickle-cell anemia, or cancer.

Knowing which missense mutations are benign and which are pathogenic has been a major challenge for researchers, with only 0.1 percent of these mutations having been clinically classified.

The creation of the AlphaMissense AI model has enabled Deepmind’s team of researchers and scientists to build upon the medical community’s decades of research and categorize 89 percent of the 71 million missense mutations as either likely pathogenic or likely benign.

Traditional experiments to uncover disease-causing mutations are expensive and laborious — every protein is unique and each experiment has to be designed separately, which can take months. By using AI predictions, such as those produced by AlphaMissense, researchers can get a preview of results for thousands of proteins at a time, which can help to prioritize resources and accelerate more complex studies.

 

Identifying Optimal Drug Combinations

Meta

Meta and Helmholtz Zentrum München have introduced an open-sourced model called Compositional Perturbation Autoencoder (CPA) that is designed to provide pharmaceutical labs, academic researchers, and biologists with AI-powered tools to help dramatically accelerate the process of identifying optimal combinations of drugs and other interventions that could ultimately lead to better treatments for complex diseases like cancer and novel diseases like COVID-19.

CPA paves the path for entirely new opportunities in developing drug treatments. In the future, it could not only speed up drug repurposing research but also, one day, make treatments much more personalized and tailored to individual cell responses, one of the most active challenges in the future of medicine to date.

 

Using Generative AI to Help Bring Drugs to Market

Bayer and Google

Bayer Pharmaceuticals, in partnership with Google, is exploring how generative AI solutions can help bring drugs to market. Generative AI can help researchers more easily access, identify, and correlate data, mine large troves of research data for possible connections, and even automate the time-intensive tasks of drafting clinical trial communications and helping translate them into different languages.

This work builds on Google’s ongoing collaboration with Bayer to accelerate drug discovery with high-performance computing power, which includes efforts to run Bayer’s large quantum chemistry calculations at scale with Google Cloud Tensor Processing Units (TPUs).

 

Clinical Trial Parser Repository

Meta

Meta AI has developed a library of AI models and data to help transform clinical trial eligibility criteria into a machine-readable format. Using this library, trials can be easily searched by their eligibility requirements, making it easier for developers and researchers to build tools that determine trial eligibility. This work will help communities provide better ways for patients from all backgrounds to access clinical trials.

 

Increasing Efficiency Across the Medical Field

NetApp

Building an AI-ready infrastructure in the highly regulated healthcare environment is anything but straightforward. For AI to thrive, data must flow swiftly and securely from diagnostic solutions at the edge, throughout clinical applications, and to cloud environments.

Given all the competing forces in medical progress, every organization must push innovation to the limits to achieve the Quadruple Aim.

NetApp is helping healthcare organizations solve performance and security challenges by providing AI-driven solutions that remove bottlenecks at the edge, core, and cloud to enable more efficient data collection, faster AI workloads, and smoother cloud integration. NetApp’s AI solutions are removing data silos to enable real-time diagnosis, speed the development of new drug treatments, and streamline administration.