Adoption of artificial intelligence (AI) has transformed every industry and sector. But perhaps nowhere can AI’s implementation make such a substantial difference than in healthcare, which can affect the lives and livelihoods of hundreds of millions of Americans.

Dr. Joe Alexander, distinguished scientist in the Medical and Health Informatics Lab at NTT Research recognizes the impact AI has in the field of healthcare.

“AI can be used to provide more individualized and effective patient care: right patient, right treatment, right time. The overall impact would be a reduction in missed diagnoses, misallocated treatments and associated excess resource utilizations.”

These are some of the ways that AI is being used in healthcare:

Remote patient monitoring and trackers

The ability to remotely monitor a patient’s room can reduce personnel, resulting in cost savings.

“Many hospitals have to use human ‘sitters’ to monitor patients that are high-fall risks to mitigate any additional injury and ensure the safety of those patients,” says Ryan Zatolokin, senior technologist at Axis Communications.

And the pandemic has exacerbated an already problematic situation among hospitals with a staff shortage.

“However, by training AI on every possible stage of a patient in bed – covers on or off, patient in the bed versus on the floor – the computer is able to recognize whether a person has fallen or not.”

He explains that AI can build the algorithm to detect when a patient is trying to get out of the bed and then alert a healthcare worker, who can hopefully intervene before the patient falls.

AI can also be used to assist patients when they’re at home. “In terms of reducing costs related to managed healthcare, currently most of the treatment recommendations are based on the previous prescription record and an assumption that the patient has followed the earlier recommendations,” says Dr. Manjeet Rege, director of the Center for Applied Artificial Intelligence at the University of St. Thomas in St. Paul, MN.

“With computer vision and natural language processing (NLP) technologies in AI, you can have digital assistants reminding patients to take medications at the appropriate times.”

In fact, he says the smart assistant can even monitor patients to ensure they do take the medication after the reminder. (Read: How Big Data Can Revolutionize Home Healthcare.)

“Today, healthcare is more prescriptive than preventive: we fix our cars before they break down on the freeway, and similarly, with AI, we can proactively prevent a future negative outcome,” Rege says.

Identifying Abnormalities and Predicting Risks

Another healthcare use for AI involves algorithms that identify physical abnormalities in a patient. Zatolokin says this can better inform a physician’s diagnosis.

“AI can build an algorithm based on tens of thousands of images of a specific object, test result, or body part. Based on that knowledge, it can be trained to identify differences and patterns, and it provides physicians with an extra set of accurate eyes.”

So how would this work? “For example, when a physician is looking to diagnose a specific eye disease, such as glaucoma, AI can be trained on all possible images of a healthy eye and also eyes with abnormalities, to determine the difference between glaucoma or other eye infections or diseases.”

As a result, there may be small changes in a patient’s eye that a doctor may not notice in the early stages. However, Zatolokin says these changes can be picked up by AI, allowing the doctor to diagnose problems earlier. (Read: What Do Patients Want From Health Care Technology?)

In addition to detecting diseases early, AI can also predict the probability of developing a certain disease or medical condition.

“AI is used in healthcare to assess patient-specific risk estimates of disease and to measure the risk factors associated with the disease, as well as to optimize processes that affect the accuracy of diagnostic tests,” says Mo Abdolell, founder and CEO of Densitas. His company uses machine learning (ML) and deep learning to identify women at higher risk of breast cancer based on patient-specific risk estimates derived from mammography exams (X-ray), so that screening protocols can be tailored to each patient with more frequent follow up screening exams or with adjunctive imaging like an ultrasound or magnetic resonance imaging (MRI).

“Machine learning and deep learning are also used to measure breast cancer risk factors, such as breast density, reliably, reproducibly, and in a standardized way so risk models that incorporate these risk factors are also reliable, reproducible, and standardized when applied across populations of patients,” Abdolell says. “In addition, they can be used to identify mammograms that are of inadequate image quality and that increase the risk of a missed cancer—which can lead to delayed diagnosis and treatment, and poor prognosis.”

Developing drugs

According to data from the Organisation for Economic Co-Operation and Development (OECD), Americans spend $1,229 a year on prescription drugs. And hopefully, AI can help reduce costs in this area, explains Rajeev Dutt, CEO and founder of AI Dynamics.

“AI can be used to create multi-targeted drugs that target multiple genes using the same drugs, and it can also customize drugs and therapeutics for individualized medicine.”

In fact, he believes that out of necessity, all of the drugs in the future will either be designed by AI or have an AI element.

“Therapeutics and drug development are getting harder, there are fewer researchers, costs are going up, and there is more competition - and this creates opportunities for multi-targeted drugs, individualized medicine, and advanced diagnostics that are a fraction of the cost,” Dutt explains.

AI can provide substantial cost- and time-savings to drug manufacturers. “The use of AI can reduce the cost of drug development from $2.5 billion in 12 years to less than $1 billion in 7 years.” That’s because AI can be used at various points during the process, including compound selection, drug safety, and biomarker identification — and that’s just the beginning according to Dutt.

“AI can also be used for commercial viability, candidate selection for drug trials, evaluation of drug trial results, marketing of the drug, and assessment of safety and side-effects once on the market.”

According to Wolf Ruzicka, chairman of EastBanc Technologies, even if the cost of developing a new prescription medicine could only be reduced by 2%, it would have a huge impact.

“By applying AI algorithms that analyze data, the patterns and correlations are much easier to identify, which saves healthcare organizations time and money.”

He says these new algorithms don’t replace existing data, but instead, provide fresh insights or a second opinion, which eliminates the need for clinical research. (Read: Top 20 AI Use Cases: Artificial Intelligence in Healthcare.)

“Regression analysis can help to find a ‘missed’ correlation between different elements of the dataset to help, for example, to uncover who would react the most or the opposite, find the limitations of who would be more tolerant to the treatment.”

Medications are never one-size-fits-all. However, Ruzicka says it’s too expensive to just research, for example, men older than 60 years of age who have a heart condition, so everyone gets lumped together.

“By capturing specific data points, such as gender, race, age, diet, activity level, health condition, and more, AI can run through the test results and identify patterns humans are unable to uncover,” he explains. “For instance, a medication that doesn't seem to lead to any significant results overall may work great for female Caucasian patients between 20 and 40 with high blood pressure who aren't pregnant.”

He points out that Viagra was originally developed for cardiovascular problems and then turned out to be useful in another area. “There are likely hundreds of medications on the market today or that were dropped along the way that could be the Viagra of COVID-19, cancer, or a rare disease with little research,” Ruzicka says.

Automating tedious chores

Companies in a plethora of sectors are using AI to handle mundane and repetitive tasks, and healthcare is no exception. Sean Lane, CEO of Olive sees many places AI could better the administrative work in hospitals:

“These tasks include revenue cycle, IT, supply chain and HR – think benefit and verification discovery, prior authorization management, invoice processing, and even scheduling. These time-consuming tasks and processes are prone to human error, devouring capacity and resources while driving up costs.”

According to Lane, data-heavy administrative tasks make up almost one-third of all healthcare costs, and he says that AI can not only help them save money, but also allow them to focus on higher-value tasks.(Read: Advancing Standards of Care Through Machine Learning.)

And there are other ways that AI can be leveraged to cut costs and save money in a hospital setting. Shash Anand, VP of product strategy at SOTI explains:

“For example, Natural Language Processing can be used to gather voice data and automatically convert it to text, inputting the data into AI systems to understand questions asked and symptoms described by patients. The data can also be automatically added to a patients’ medical record in an appropriate format, which saves doctors the time required to enter the data manually.”

And as a bonus, he says doctors can spend more time on quality interactions with patients.

“This data can further be leveraged for chatbots, as an example, to help patients comprehend complex medical language should they not understand terminology that is being discussed or if they have follow-up questions after the appointment,” Anand explains.

Handling Logistics

AI can also serve a logistician function in healthcare by identifying patterns in hospital layouts.

Says Zatolokin: “Busy hospital hallways can get congested, from patients waiting on gurneys in the corridors, to departments being overcrowded due to an incident. While there is a lot going on with patients, there is even more happening behind the scenes to ensure doctors and nurses have the right supplies and floors are stocked with what they need.”

He says AI can be used to quickly move the right resources to the right place in a short amount of time. (Read: Can IoT Improve Supply Chain Management in Healthcare?)

“For example, AI can help determine the difference between a gurney with or without a person, or a hospital bed versus a group of boxes.” Also, when a hospital employee needs to quickly transfer a cart of supplies from the basement to another floor, Zatolokin says an AI-created algorithm can quickly analyze the time of day and highly populated areas to determine the best route.

With AI becoming ever more integrated into our daily lives, the possibilities are exciting. Experts are watching the healthcare field as human capital is freed from tedious or repetitive jobs, while machine learning and advancing technologies are utilized to recognize and refine the things humans cannot.