By Clifton Roberts, Global Director, International Trade Group
Oftentimes, when people think of AI, they think of a cold, logical, hyper-intelligent program with no empathy — not a mechanism to augment personalized human interaction. Yet this is what empathetic AI has been designed to do in the healthcare industry: perform complex tasks and analyses in order to free up time for health practitioners to interact directly with patients and form deep human connections. For years, Intel has been collaborating with healthcare experts and medical researchers to deploy empathetic AI capable of helping tackle many of the biggest healthcare challenges we face, including treating cancer and HIV.
As part of this effort, Intel in collaboration with Enzolytics, a developer of therapies for infectious diseases around the world, to publish a whitepaper, “Optimizing Empathetic AI to Cure Deadly Diseases.” This whitepaper outlines some of the recent medical advances Intel has been part of, including using deep learning to improve assessment of radiology images and subsequent diagnoses, to gain insight into tagging of protein sequences, to improve the accuracy and speed of screenings for cancer, and to map brain activity to develop advanced neurocognitive therapies and decode human thought.
The whitepaper also outlines how Enzolytics has used AI and advanced medical technologies in the search for an elusive HIV vaccine. By conducting an analysis with AI, researchers were able to validate research into “immutable” sites (sections of HIV’s RNA sequence that remain constant even as the virus mutates over time). With these immutable sites, doctors should be able to develop antibody treatments that attack these specific sites and neutralize (or at least lessen the impact of) the HIV virus. Though this is a long and complex process, AI has helped accelerate the work by providing high-speed analyses of large amounts of data.
For this kind of work to be possible on a broader scale, we believe specific investments are necessary to ensure the accuracy and efficacy of empathetic AI in healthcare. Here are some of our recommendations:
- Deepening public-private partnerships: This past year has taught us how important effective partnerships between the U.S. government and private companies are when addressing public health issues and keeping the general population healthy. With the U.S. government currently purchasing half of all healthcare services in the country, we foresee public-private partnerships being critical to maximizing the good that empathetic AI can have in our society.
- Implementing standards and frameworks: When you work with the vast amounts of data needed for AI, that data must be arranged into standardized sets and approached with standardized procedures to ensure accuracy and usability. That’s why we should foster secure data sharing between public and private entities, as well as create public databases of freely accessible and anonymized data for secure analysis without risk to patients or consumers.
- Minimizing bias: Like all forms of AI, empathetic AI suffers when the programmers who develop them transfer their bias to the AI or the datasets on which the AI relies. We see the need for larger and more diverse datasets, as well as for more diversity in the field of AI development and research. With more perspectives at the table at every stage of the development process, AI will be less biased and more reliable, as will its datasets.
To read the rest of our recommendations, check out the full whitepaper, “Optimizing Empathetic AI to Cure Deadly Diseases.”