Here’s How AI is Helping Physicians
An insight into the use of VR simulations in supporting mental health management.
Technology has swiftly made its way into the healthcare industry, including mental health.
Digital tools and technology have already paved their path in what we call mental health technology — supporting the prevention, diagnosis, treatment, and management of mental health disorders. It started with mobile applications, online therapy platforms, and telemedicine services and now VR simulations have joined the league.
You heard that right. I’m talking about VR — Virtual Reality.
Reducing Physician Burnout
Physician burnout is a prevalent issue impacting physicians, the medical community, and patients. An article in Mayo Clinic Proceedings highlighted that up to 62.8% of U.S. physicians experienced burnout in 2021.
One of the ‘root causes’ of physician burnout stems from the pile of non-productive tasks physicians are bound to do during consultations. A 2018 study conducted by Stanford Medicine revealed that 74% of primary care physicians spend more time on EMR than with patients.
Instead of focusing on valuable tasks like consultancy, doctors become data entry clerks rummaging through piles of files and documents in the EMR. These factors lead to career satisfaction among physicians and impact their physical and mental health.
This is where the most significant advantage of AI comes in – and that is its ability to automate repetitive and non-productive tasks so physicians can focus more on consultations.
AI Tools to Reduce Physician Burnout
Here are examples of some AI tools being developed and tested to help reduce physician burnout by taking charge of mundane tasks:
Nabla is an AI tool that records real-time physician-doctor conversations, summarizes them, and generates them as reports. This liberates doctors from documenting and writing prescriptions to focus more on consultation.
Another example is a co-pilot tool, Regard. It goes a step ahead by mining the medical record, using it to assist physicians in diagnoses, and then drafting clinical notes that the physicians can review and sign.
Reducing Cognitive Burden
While administrative burden results from the hassle of dealing with EMR and prescriptions, cognitive burden results from the mental strain physicians go through in understanding, processing, and making clinical decisions while keeping up with the ever-evolving medical science every day.
Here’s how:
Medical science is forever changing, with new treatment options emerging every other day, resulting in the exponential growth of medical data. Even the most experienced physicians must stay abreast of the newest updates but struggle to keep up with all the medical information to make the right decisions.
A study by Tom Lawry discussed that the rapid growth of medical data doubles medical knowledge every 72 days, placing an intense cognitive burden on physicians. AI can assist doctors in managing this overwhelming influx of data, easing their cognitive load.
A study published in the Journal of Medical Internet Computing (JMIR) explores semantic search and natural language processing (NLP) to reduce cognitive load in EHR systems.
Findings from this study propose that employing AI tech here can assist doctors in handling the surging amount of clinical information, helping them make informed decisions for patient care.
AI Tool To Reduce Cognitive Burden
The co-pilot AI tool, Regard helps physicians reduce cognitive burden by using patient data to make diagnoses and draft clinical notes for review and signature. In a case study, Regard was found to reduce physician burnout by 50%.
Assisting with Patient Portal Messages
Responding to patient messages is another mundane task that drains physicians and takes their time. Because of time constraints, physicians give short answers to patient queries.
Again, AI can help physicians by extracting information from the patient’s EMRs to draft a long, detailed response to the patient, which the physician can review and edit before sending.
This allows physicians to respond quickly to patient messages, save time, and improve the quality and efficiency of medical care.
Generative AI Tools to Assist With Patient Portal Messages
Several AI tools and AI chatbots are being studied to be used to respond to patient requests. Let’s have a look at a few:
Doctors at UC San Diego Health are using an AI-assisted tool to respond to patient queries on the MyChart Patient Portal. It extracts information from the patient’s EMRs and drafts initial responses, allowing doctors to focus more on direct patient care.
Here’s a case scenario explaining how it works:
Patients concerned about specific findings from their bloodwork ask their queries on the portal. The AI chatbot drafts a response explaining whether or not the reading is within normal range and then guides the patients on the next steps for making an appointment. Once the response is generated, it is then reviewed by the physician for accuracy.
Several other health systems, like Ochsner Health and Scripps Health, are developing similar AI chatbots to assist physicians in responding to patient portal messages.
Clinical Decision Support for Chronic Care Management
Clinical Decision Support, or CDS, is an important tool in healthcare decision-making. Healthcare providers routinely use a Clinical Decision Support System (CDSS) in their clinical practice. Like other areas of healthcare tech, AI hasn’t spared CDSS and is now being integrated into the system to manage the growing volume of data.
AI-Assisted Clinical Decision Support System (CDSS)
Scientists and health technologists are now studying AI-assisted CDSS for their potential to enhance diagnostic accuracy, make informed decisions, and assist physicians.
The MIT Technology Review shared an insight on one such tool – Sepsis Watch used to help doctors reduce sepsis-induced patient deaths. Sepsis Watch was developed as a deep learning tool at Duke Institute of Health Innovation to help doctors with the early detection of sepsis.
Sepsis is a life-threatening condition that leads to full body inflammation and quickly progresses to septic shock and end-organ failure. Because of its vague symptoms, it is often mistakenly diagnosed for other conditions, putting patients’ lives at risk in emergency care.
The Sepsis Watch was designed to help physicians with early detection by monitoring patients frequently for any visible signs of developing sepsis. Through Sepsis Watch, doctors at the Duke University Hospital could identify patients at medium and high risk and treat them promptly.
Compiling Clinical Summaries
All healthcare providers rely on clinical summaries to diagnose and treat patients and keep track of their disease and medication history. However, physicians and healthcare systems struggle with ‘keeping up’ with the growing medical records of their patients.
This is again where generative AI comes in as a savior. Halmet is one such AI-generated tool developed to summarize the text. It can extract information from copied text and PDF files to create a clean, summarized text. AI tools like these help physicians compile clinical summaries, improve their decision-making process, reduce review time, improve accessibility, and detect errors.
Another innovative AI tool worthy of discussion is HealthScribe, which was developed by Amazon Web Services (AWS). It helps physicians by simplifying clinical documentation. Using speech recognition and AI, HealthScribe listens to patient-doctor conversations, extracts essential information, and generates summaries for electronic health records.
Wrap Up
AI is rapidly evolving in the healthcare industry, helping physicians win the battle against time-consuming tasks and information overload. From alleviating physician burnout to helping clinical decision-making, the potential impact of AI is palpable.
What makes AI even better is its ability to automate mundane tasks while still keeping the overall healthcare experience “human-friendly.” AI tools allow physicians to build better bonds with their patients by liberating them from trivial tasks like documentation and data overload.