How Healthcare Industry Is Using NVIDIA AI to Better Meet Patients’ Needs

AI has the potential to transform healthcare. And in few places is that potential greater than in fast-growing China, home to one of the world’s largest elderly populations.

As the country’s healthcare industry works to address the needs of this aging demographic, technology is playing a central role — especially in medical imaging and analytics of electronic data records.

The government is backing this effort with its smart health initiatives, which is encouraging hundreds of AI healthcare startups and large industry players to embrace AI as the foundation for the industry’s future.

Ping An, BGI, iCarbonX Supercharge Data Analytics

To get a sense of the scale of China’s challenges, consider the insurance giant Ping An, which has nearly 180 million customers.

To gain insights on problems like fraud detection or to predict disease in populations, Ping An’s data science team uses a popular open-source machine library, called scikit-learn, and two machine learning algorithms: principal component analysis (PCA) and density-based spatial clustering of applications with noise (DBSCAN).

Ping An recently tried out RAPIDS, a new open-source, GPU-accelerated platform for large-scale data analytics and machine learning. For the first time, it allows data scientists to run data science pipelines on GPUs — and realize massive speedups in the processing time for their datasets.

When Ping An used RAPIDS and GPU-accelerated PCA and DBSCAN, they clocked 80x speedups — from days down to hours — on their workflow, including data loading and training time. This helps them develop proactive predictions and improved precaution and prevention plans.

Similarly, China’s biggest genetics company, BGI, has a sea of data — more than 1PB, housed in a data warehouse it calls the Knowledge Base. BGI uses a machine learning algorithm called XGBoost to classify targetable peptides that can personalize immunotherapy for cancer patients.

By running the RAPIDS platform on an NVIDIA DGX-1 AI supercomputer, the BGI team sped up analysis by 17x and expanded its analysis to millions of peptide candidates.

Another company leading the way is iCarbonX, which focuses on digital health. One of its solutions uses increasingly popular digital physiology, genomics, metabolism and metagenomics data to study the microbiome.

By applying machine learning to correlate microbiome features and Type-2 diabetes, iCarbonX can deliver personalized consumer health services such as diet suggestions or treatment planning. RAPIDS deployed on Tencent Cloud P40 servers resulted in a 6x speedup out of the gate for the company.

Top Medical-Imaging Companies Enhance Efforts with GPUs

Medical imaging is the earliest beneficiary of AI in healthcare, riding on the coattails of the consumer internet boom of images and video. However, while 70 percent of medical-imaging research is now based on deep learning, few algorithms have made their way into clinical deployment.

The reason is medical imaging AI tends to be sensitive to things like patient demographics, age of the imaging instruments and settings of those instruments when the images are acquired. Variables like these can negatively affect the accuracy of AI.

For this reason, AI applications needs to be developed locally, and that’s just what two of China’s leading medical imaging companies are doing.

United Imaging Intelligence, one of the leading medical imaging AI companies, is building out AI infrastructure based on NVIDIA DGX systems to develop its full-stack medical imaging AI software, called uAI. The company’s goal is to develop superior AI software and products for a full range of medical imaging workflows, including image collection, disease screening and treatment solutions.

Likewise, Infervision, founded in 2015, is one of the top medical imaging startups in China and has collaborated with several hospitals in Asia, Europe and North America. Infervision is adopting the NVIDIA Clara platform to power its next-generation AI imaging cluster. Its InferRead solution deployed GPUs in AI systems for computer-aided diagnosis at hundreds of the world’s top hospitals.

By embracing innovation, China and other countries have the opportunity to build an AI-first healthcare industry that can benefit all of their people.

The post How Healthcare Industry Is Using NVIDIA AI to Better Meet Patients’ Needs appeared first on The Official NVIDIA Blog.