AI Blazes A Path For Faster, More Accurate Care Delivery

By Enakshi Singh, Senior Product Specialist, SAP Health

With applications ranging from mining medical images for disease to managing patients with chatbots, artificial intelligence platforms have learned how to derive value from data.

Consulting a colleague is standard practice among physicians when they are unsure of a patient issue, such as a skin disorder or other anomaly, requires a biopsy or further testing. While getting that second opinion is important, the validity is still limited to the second physician’s training and experience. That’s where artificial intelligence (AI) is poised to make a breakthrough, one that can result in better health delivery for all.

Using large pools of validated datasets, AI can discover and learn, turning data into usable — and actionable — knowledge. Not just relegated to surgical robots and machinery, AI platforms coupled with in-memory computing technology are proving to have enormous potential in many areas of healthcare, including pharmacology, imaging/radiology, ophthalmology, and oncology.

AI is coming into play in many different areas of healthcare, including in the physician’s office, where it can rapidly scan images and provide data that leads to a more accurate and timely diagnosis of difficult diseases, such as skin cancer and chronic ailments. AI is also at the front lines of patient care, as cognitive virtual agents — medical chatbots or “chat nurses” — use embedded AI technology to augment patient management, helping to determine whether a sick patient needs immediate care.

Discovering Patterns

The key enabler to the success of AI in healthcare is sharable data. As healthcare organizations embark on more digital transformation initiatives, vast pools of electronic data are becoming available in the public health ecosystem. The many sources of patient data include clinical trials, electronic health records (EHRs), pharmaceutical, hi-resolution images, and genomic profiles, all of which can be useful for care delivery.

These enormous data pools need to become more open and sharable in order to create a comprehensive reservoir of structured and unstructured medical information. Using algorithms, AI systems can discover patterns in these data pools. The AI system rapidly learns how to hone in on key information, develop hypotheses, and refine its answers over time. The system is trained to figure out a problem, rather than have the answers programmed in, allowing the cognitive platform to increase in value over time.

Fast and Accurate Clinical Support

One of the key benefits provided by AI to physicians is to help with clinical decision support. Doctors are challenged with every patient to solve a complex data problem, coming up with a diagnosis based on symptoms, patient history, clinical tests, and medical images. This is where AI has already made an impact: in evaluating imaging data.

According to the Radiological Society of North American (RSNA), approximately 1 billion radiologic imaging examinations are performed worldwide annually, and most of the resulting images are interpreted by radiologists. While a single radiologist may have evaluated hundreds of images, a typical AI system can distill actionable insights from billions of clinical cases. It can leverage collective intelligence from each image in a matter of seconds. These AI systems have astounding speed. For example, Enlitic, a San Francisco-based deep learning company, has an AI platform that can interpret a medical image in milliseconds — up to 10,000 times faster than the average radiologist.

Another benefit of AI platforms is accuracy. The National Institute of Medicine estimates that diagnostic errors affect 12 million Americans each year. An AI system can more accurately and efficiently classify tumors, which is particularly important in cancer diagnosis and treatment. The Enlitic AI technology can judge the malignancy of chest nodules from a lung cancer screening 50 percent more accurately than an expert panel of radiologists.

Improving Patient Engagement

In addition to helping doctors, AI is also helping patients, as a new wave of automated patient management tools is emerging. The goal is to improve patient engagement after they leave the hospital or facilitate chronic disease management, such as diabetes. Multiple studies have shown that healthcare providers gain cost savings and improved outcomes by staying in touch with patients after they are released from a hospital or after a procedure.

One promising AI-based technology is a new model of interaction, medical chatbots. By using natural language tools, AI developers working in tandem with clinical experts are building modules to augment patient contact. Deployed into health platforms, the AI chatbots can access large repositories of public chat scripts and mimic human conversations, with the ability to engage patients empathetically. Since AI platforms can do complex image recognition, a patient could use this type of system to upload an injury photo or add a handwritten note, and the technology could analyze and factor the data into its diagnostic process.

With AI analysis, a chatbot that can access a patient’s health record, knows the list of prescribed medications, and interprets current symptoms could determine whether a patient needs to get immediate treatment for an ailment. Ideally, the AI system could interface with an online appointment system, helping to get the neediest patients quickly to a physician. Companies such as Sensely, YourMD, MedWhat, and Babylon Health are just a few examples of platforms using medical AI chatbots.

Getting Value From Health Data

While integration of AI tools and cognitive computing in healthcare has been slow, it is on the move in 2018. AI platforms with access to clean, well-governed data sets can work at incredible speeds and be trained to produce accurate results. AI is set to be the smart assistant for physicians as well as patients, one that can find patterns and reveal predictive data, helping to determine the risk of disease or the best possible treatment in milliseconds. Most importantly, AI is one of the tools poised to transform how healthcare organizations can get value out of their data, making it an intelligent choice that can benefit all.

About The Author

Enakshi Singh is a member of the Healthcare Development team, with a focus on product management in genomics & healthcare. She works with multidisciplinary teams at SAP and collaborates with renowned genomic researchers to develop software solutions for real-time analyses of large-scale biological, lifestyle and clinical data. She brings together a unique blend of strong academic training along with hands-on experience in working closely with cross-functional teams. Enakshi received her M.Sc. in Neuroscience at McMaster University in Ontario, Canada. Her Masters’ thesis focused on the developmental distribution of the NMDA receptor in the auditory brainstem of rat. She also earned a BSc. in Psychology, Neuroscience and Behavior at McMaster University. In her spare time, Enakshi enjoys staying up to date on the latest scientific literature and technological advances in healthcare.