AI in Action: Healthcare
This is the first in a series of articles highlighting the many applications of Artificial Intelligence
Healthcare is never far from the headlines these days. The tales of astronomical costs and service deficiencies are balanced by the latest reports of new treatments and drug breakthroughs. The first year of the baby boom generation is now 72, so the pressure on healthcare services is rising to unprecedented levels.
Healthcare is crying out for innovation, and AI is here to help. These are just a few of the ways that AI can provide assistance.
Customer Service & Administration
Healthcare is drowning in paperwork and struggling to provide acceptable levels of service. From managing emergency rooms to scheduling appointments to completing paperwork — there are many ways that AI can streamline healthcare to improve service and contain costs.
Chatbots are a natural application for healthcare. They can answer basic questions for patients and families. They can also collect information before a chat caller is passed to a live agent. Chatbots can also act as a virtual assistant for patients — keeping track of appointments, scheduling reminders for call backs, and helping to fill out paperwork.
Chatbots can also act as real-time assistants for health hotlines, particularly hotlines like a kids hotline or a suicide prevention line that are often staffed by volunteers. A chatbot that offers suggestions to volunteers can improve the level of service, reduce training times, and make the job more manageable for volunteers.
Chatbots can also act as virtual assistants for physicians — transcribing notes, managing calendars, and sorting and prioritizing documents. They also offer a new, often more painless way to fill out forms. A chatbot can take an onerous task like completing an insurance form, and help the patient finish the form simply by asking a series of questions. Many people prefer the conversational interface of a chatbot to filling out a form on their own.
Diagnosis and Anticipation of Health Issues
A primary function of artificial intelligence, particularly neural networks, is building better predictive models. There are countless applications of predictive models in healthcare. Some examples would include predicting blood sugar levels in diabetics, early identification of Alzheimer’s based on speech patterns, or anticipating seizures based on EEG data. Any health-related issue with a rich set of data has potential for building an effective predictive model via AI.
Medical imaging is a excellent source of data for any type of predictive model. Combining imaging data with other sources can yield a predictive system. With a new NFL season upon us, CTE risk is top of mind. There is no predictive model for CTE risk at this time, but the components to create a model are coming into place. Combining medical imaging (CT, MRI, x-ray, etc.) data with physiological and genetic data sets could produce a solution that will give some guidance as to the individual risk for each player. There may come a day when a prospective football player will be armed with an estimate of their risk for permanent impairment from playing football.
Soon AI solutions will be able to interpret medical imaging data and provide radiologists with a preliminary diagnosis for review. This will save time and provide radiologists with an instant “second opinion”. This is one of many ways that AI will be able to act as an able assistant for practitioners, rather than replace them.
Drug Design & Discovery
The preliminary identification and fine tuning of new drugs can be done via computer modeling. AI can be invaluable in shortening modeling times, and recognizing the most promising drug designs. More precise and robust drug designs can lead to shorter time to market and higher efficacy.
Medical equipment, from treatment devices to imaging systems to diagnostic and administrative solutions are already having AI solutions embedded into them. These “blackbox” applications improve accuracy, reliability, and speed. Although unseen by the public, embedded applications will most likely be the backbone of the AI evolution within healthcare.
This is just a small subset of all the ways AI will impact healthcare in the coming years. It will be a long time before robots are doing surgery unassisted, but AI is already improving outcomes in healthcare today.
Ken Tucker is an AI and Analytics business consultant.