Top Five Trends for AI Startups in Healthcare Life Sciences

Thirty percent of the world's data is being generated by the healthcare industry. Vast amounts of data, coupled with robust AI models powered by strong compute, are creating a new era of healthcare. Disruptive medical research breakthroughs are advancing the discovery and practice of medicine.?

Healthcare startups saw a record high of funding of $30.7B in Q3, 2021, a 41% increase YoY, according to CB Insights. NVIDIA Inception, a program designed to nurture startups revolutionizing industries with advancements in AI and data science AR/VR, and beyond, is now nurturing more than 1,400 healthcare life science startups in drug discovery, pharmacology, genomics, 3D medical imaging, intelligent medical instruments, and smart hospitals. Many of them will be presenting at NVIDIA GTC , the No. 1 AI conference, which begins on Monday November 8, with Jensen’s keynote going live on November 9th.?

To help you navigate the conference, here are the top five trends we’re seeing among healthcare AI startups:?

I) Digital Biology Revolution

The traditional drug discovery process is following Eroom’s law, which is the observation that drug discovery is becoming slower and more expensive over time, despite improvements in technology such as high-throughput screening, biotechnology, combinatorial chemistry, and computational drug design. The amortized cost per approved drug is $2.5B with an aggregated R&D success rate of 5%. Scientists are turning to AI and simulation to accelerate this process. The decreasing cost of sequencing has led to a proliferation of genetic data. AlphaFold, a solution to a 50-year-old grand challenge in biology, released 130 million protein structures online for protein structure prediction. RoseTTAfold enables the protein structure to be calculated in just 10 minutes on a gaming computer. Deep learning generative models are also used to generate synthetic data similar to real observation to speed drug discovery processes.?

These forces form the perfect storm for digital biology, a field of research where powerful simulation and AI software are used to understand the basic function of human life. NVIDIA is advancing the digital biology revolution by accelerating every phase of drug discovery-from accelerating sequencing, to determining the structures of proteins, to accelerating image processing in microscopy, to accelerating docking and simulation calculations, to optimizing chemoinformatics approaches.??

Using computers to simulate experiments to identify drugs with therapeutic benefits is key to industry breakthroughs. Just a year ago, it took a mere 12 hours to screen 1.4 billion compounds against a protein with high accuracy on Summit, Oak Ridge National Lab’s (ORNL) supercomputer with 27,648 GPUs. It was a revolutionary computing breakthrough. When you combine data, computing, AI, and simulation for next-generation therapeutics, some of the most pivotal innovations of the 21st century are underway. Check out the latest breakthroughs from our life science AI startups, including Entos, Qubit, Cyclica, and Immunai , at GTC.?

II) Federated Learning (as a Platform?)

While we are sitting on a plethora of data, there’s an increasing concern of transmitting private patient data in healthcare. Federated learning (FL) has since emerged as a robust way to train your AI model without exchanging the data itself and preserving privacy.

Since the launch of such techniques a few years ago, Nature Medicine published a paper on “Federated Learning for Predicting Clinical Outcomes in Patients with COVID-19,” where 20 institutes across the globe used federated learning to train a model that predicts the future oxygen requirements of patients with a potential COVID-19 infection using inputs of vital signs, laboratory data, and chest X-rays.?

FL has now expanded across use cases in medical imaging, biology, and pharma. First mover advantage will belong to those who figure out the use cases, the business model, and the network effect. FL startup Rhino Health will share how they develop AI in clinical settings with their federated learning solution working with leading networks/academic medical centers and industry developers.?

III) Medical AI at the Edge?

Storing data locally, preserving patient privacy, and pushing network computation to edge devices are collectively an increasing trend. When a matter of seconds could determine the outcome for a patient, it’s vital to have edge AI -- the combination of both edge computing and AI -- moving AI to where the data generation and computation take place for better patient care. Implementation of healthcare AI is expected to grow at an annual growth rate of 41.4% from 2020 to reach $51.3 billion by 2027, while edge cloud computing is expected to grow by 34.1% between now and 2025, according to Med Tech Innovation News.?

We are at the dawn of large-scale deployment of edge AI in healthcare, delivering new use cases that did not seem possible before. With smart sensors, optimized AI models, and intelligent edge networks, hospitals can get the help they need to advance patient care, increase data security, and improve operational efficiency.?

Trend setters like Viz.ai empowers clinicians in more than 850 hospitals with AI-powered Intelligent Care Coordination; HeartFlow Analysis, an AI-powered solution to help diagnose and treat coronary artery disease, has been adopted by 500+ hospitals worldwide; The Hyperfine Swoop?, the world’s first FDA-cleared portable MRI device, is in use across more than 50 hospitals; Artisight has deployed its end-to-end IoT platform for patient monitoring across 30+ hospitals.?

IV) Enterprise-Ready Imaging AI

It’s proven that AI can ameliorate healthcare worker burnout, help healthcare workers to be more efficient, and find anomalies faster and offer better patient care. However, just 53% of AI projects make it from pilot to production, according to Gartner. Organizations spend on average $500K and months of time evaluating one AI application and even longer for integration. With the influx of AI applications, it’s even more important to have tested and certified applications that can be seamlessly integrated into hospital and clinical environments.?

NVIDIA AI Enterprise (NVAIE) is an end-to-end, cloud-native suite of AI and data analytics software, certified and supported by NVIDIA to run on VMware vSphere with Tanzu and optimized for NVIDIA-Certified Systems. NVAIE streamlines the complex process of integrating new applications into existing infrastructure to make AI ubiquitous in healthcare. This, in turn, can increase operational efficiencies and patient satisfaction. Startups are not just focusing on building AI models, but also how to deploy their solutions at scale effectively. Check out how to build smart hospitals with Mass General Brigham and AI applications Vyasa and iCAD at GTC.?

V) Healthcare Innovation In Emerging Markets

While we see tremendous advancement of AI in healthcare in developed countries, half of the world has no access to essential health services. The UN predicts that, by 2050, the world’s population will exceed 9.7 billion people, with emerging market countries accounting for over one-third of total population and having over 588 million people aged 65 or more. By 2040, emerging market countries on average are projected to increase healthcare spending as a portion of GDP by 24.4% compared to just 9.8% in developed markets over the same period.?

Considering that 90% of the world’s population is in developing countries, it’s a huge market to address. Using medical imaging as an example, 95% of patients need some form of medical imaging in their treatment. Companies in developing countries have a big advantage in developing AI-based healthcare technologies. Solutions like AI in healthcare are bringing new innovation to reach underserved communities. At GTC, Noul , a Korea-based biotechnology company that makes diagnostic platforms for malaria and blood cell morphology, has tested more than 4K clinical samples and 11 global clinical trials. DilenyTech deploys innovative AI and medical imaging solutions in Egypt, Africa, and across other continents. MinoHealth AI Labs uses AI for Biotechnology, Radiology, Oncology, Regenerative Medicine, Neuroscience, Optometry use cases in Africa.?

The top five healthcare life science trends to watch are digital biology, federated learning as a platform, medical AI at the edge, enterprise-ready software-defined medical AI devices in emerging markets. All GTC talks can be found here .