Back

DataJan 18, 2021

Six Use Cases for Artificial Intelligence in Healthcare

Shawnasty Bankovich and Stephanie Rivera

As COVID-19 continues to impact the globe, enhancing healthcare using artificial intelligence (AI) is now more crucial than ever. AI has helped revolutionize the healthcare industry, and PwC estimates the AI market for healthcare applications will erupt from $663.8 million in 2014 to $6.7 billion in 2021. This increased demand correlates with a substantial rise in complexity and abundance of data.

At Credera, we have worked with clients in various industries, including healthcare, to design, develop, and deploy AI solutions. Through these experiences, we have identified six key areas in the healthcare field where AI can have a significant impact. This blog post will highlight the potential business value of leveraging AI in each area.

1. Patient Prescreening

In patient prescreening, AI can add value even before arriving at a medical facility to seek care. Today, prescreening can take the form of questionnaires or online symptom checkers like WebMD. In addition to asking leading questions that may induce bias, a leading medical journal claims these online symptom checkers can have a wide range of accuracy. For example, in 2015, Harvard teamed up with Boston Children’s Hospital to analyze online symptom checkers’ prognoses. The results revealed significant flaws: Correct advice was provided only 58% of the time across 23 online symptom checkers. AI can provide a solution to this problem.

AI can improve patient prescreening by providing a more user-friendly experience than standard prescreening questionnaires through expanded voice- and text-based communication. Additionally, AI can learn new patterns and more accurately diagnose patients’ symptoms through machine learning. Text- or voice-based artificially intelligent chat solutions can mimic a real-life healthcare professional’s expertise, much more than a predefined, static survey. Thanks to natural language processing (where computers identify text patterns), a potential patient would explain symptoms in voice or text as they would to a physician. The patient and AI agent will ask each other questions, simulating a real-life physician and patient conversation.

AI agents could also provide the next steps in seeking care to the patient, such as visiting an emergency room, scheduling an appointment with a specific type of physician, or taking over-the-counter medication. If the patient needs medical attention, the AI agent could recommend a medical professional based on location, availability, and insurance information. With the patient’s consent, the agent may even share their records with a medical professional, allowing for an earlier, more complete diagnosis.

2. Patient Intake and Triage

Patient intake and triage can be an inefficient process and is another promising area for leveraging AI. Medical emergencies can be highly stressful situations and seeing a medical professional often involves time-consuming paperwork. Filling out this important paperwork during an emergency can be challenging, but AI can help with patient intake and triage by using chatbots or voice assistants to quickly gather necessary information from patients and determine the appropriate next steps.

Patient data is automatically digitized when using AI for intake. The digitized data will allow for more efficient scheduling algorithms to match patients to the most appropriate doctor. AI algorithms could also use digital patient intake data to determine if someone should wait to see a specialized doctor or if they need the first available care. For example, perhaps patient X came in before patient Y but, due to the nature of ailments, patient Y must be seen before patient X. In the U.S., emergency room wait times range from two hours in Nebraska to over 14 hours in Puerto Rico. Using an AI intake solution could drive down ER wait times and improve patient care.

Using AI in patient intake can provide a less stressful patient experience while most appropriately utilizing medical personnel’s time. Healthcare institutions can also reap the benefits of decreased patient data collection costs by removing the data entry required to digitize medical records.

3. Diagnosis and Medical Imaging

One of the most well-known applications of AI in healthcare is medical diagnosis. Diagnosis and medical imaging are strong use cases for AI because machines are well equipped to handle rare events such as uncommon diseases. They excel in advanced pattern recognition on a wider variety of anomalies than most human healthcare professionals can recognize.

Both medical imaging and diagnosis are very complex problems. Traditionally, they’ve used feature development and engineering (transforming data to represent or capture concepts as input for a model), but that requires a subject matter expert (SME). However, deep learning (DL), a sophisticated form of AI, is a compelling solution for difficult topics requiring human-like “thinking” such as medical imaging and diagnosis. This AI application allows medical institutions to decrease the amount of time highly skilled personnel spend studying medical images and will enable personnel to tackle more challenging problems and treat more patients.

4. Preventative Healthcare

What if we could take steps to prevent a disease before it manifested and turn the outlook from reactionary to preventative? Preventative care is any service or procedure that defends against future health issues. Today preventive care is limited to procedures such as routine physicals, vaccines, dental cleanings, etc. There is an opportunity to harness the predictive capability of AI applications to upgrade current preventative care.

AI applications are robust in predictive maintenance in many industries, such as manufacturing, automotive, and aerospace, often relying on sensors. Similarly, in healthcare, wearable fitness and medical devices such as Apple Watch, Fitbit, Garmin, and others can act as sensors that support preventative care. Natively these devices support data such as heart rate, physical activity, nutrition, VO2, and sleep. With add-ons, these devices can now even track blood pressure, blood sugar, and weight. The data collected from these wearable devices could revolutionize preventative healthcare by forecasting health issues before they occur. The large amount of data collected from these devices could be used to design analytical models that predict specific diagnoses. Real-time wearable sensors could send their data through these models, alerting users when they have reached a threshold for requiring physician preventative care. Their wearable data could be shared with the physician for a more in-depth medical understanding, driving a higher quality of care and reducing the time spent on patient medical symptoms and history.

5. Drug Discovery

COVID-19 has shed light on the vaccine discovery process. The world has been working toward a vaccine for a year, and while significant progress is being made, the vaccine and drug discovery processes are very manual, expensive, time-consuming, and often not fruitful.

As a result, it is becoming increasingly popular for drug makers to turn to AI solutions such as DL to develop and test new drugs. Developing new drugs requires processing large amounts of data due to the sheer number of possibilities of chemical combinations (or what is known as the chemical space). AI addresses this challenge by harnessing its ability to perform incredibly well with large amounts of data. This ability produces higher numbers of approved drugs in a faster, cost-cutting process with higher research and development efficiency. In fact, Moderna, one of the companies to produce a COVID-19 vaccine, has been known to utilize AI in their vaccine discovery process.

6. Optimized Standard of Treatment

The Center for Disease Control (CDC) cites that over 85% of physicians use electronic medical record systems. Digitized medical records open many opportunities for AI, such as optimizing the standard treatment (ST). The ST, also known as the standard of care, is the medical treatment that is normally provided to people with a given condition. An ST guide may be a compiled list of best practices vetted from SMEs for medical care based on a patients’ symptoms and medical conditions. However, medical advances and understanding change frequently. AI could keep the ST guide up to date. Additionally, Bayesian learning methods (a type of AI) can be used to integrate current patient medical records and treatment information into the ST given by the SME, evolving into an optimized standard of care. This optimized ST will be continually learning from current treatment information, allowing the AI to learn and grow as new treatments are developed.

An optimized ST plan assures that each patient receives the most current and informed medical care for their conditions, which is clearly a great benefit of AI for patients. A less obvious benefit is the reduction in mistakes, especially when faced with extremely rare diseases. Doctors would have the best information to assist in decision-making. Therefore, hospital ratings, which is how hospitals are compared, would be higher for those using optimized ST plans.

Moving Forward

Healthcare is a discipline with challenging, high-stakes problems. With the advent of AI, healthcare systems and processes can be improved or streamlined; these include:

  • Patient prescreening

  • Patient intake and triage

  • AI diagnosis and medical imaging

  • Data-driven preventive care

  • AI-assisted drug discovery

  • Optimized standard of treatment

Credera has helped our healthcare clients develop and integrate AI into their operations to better leverage technology toward growth and improved patient care. If you’re interested in achieving best in class AI capabilities within your organization, we’d love to help! Please reach out to us at findoutmore@credera.com.

Have a Question?

Please complete the Captcha