Medical Coding and AI: The Future of EHRs
Search For Schools
When you click on a sponsoring school or program advertised on our site, or fill out a form to request information from a sponsoring school, we may earn a commission. View our advertising disclosure for more details.
“Some of the tasks [healthcare workers] need to do are so administrative and so repetitive. So for every task that is repetitive, AI could definitely bring a lot of help and make a huge impact on how they are experiencing their work.”
Michal Tzuchman Katz, MD, CEO of Kahun (AI Company)
We are lucky to live in the golden age of medicine. Within the last 100 years, major advancements have been made to eradicate previously fatal diseases, vastly reduced maternal and infant mortality rates, and bumped up average life expectancies by high margins. And there are still many burgeoning avenues of research and medical innovation to anticipate.
However, these advancements can only benefit the public if there is a functional system in place to deliver care to everyone when they need it, which is why one of the greatest limitations and most precious resources in healthcare is time.
This is true in emergency settings, where studies show that patients who waited eight to 12 hours have a 10 percent higher likelihood of mortality than patients who waited six hours. But it’s also true in preventative care.
Studies have also shown that when doctors and patients are less rushed during their appointments, patient safety, medication adherence, and increased patient satisfaction across the healthcare sector and costs of medical malpractice decrease.
Short staffing and long wait times are universal problems in healthcare systems worldwide. Not only do these issues contribute to poor patient health outcomes, they also put pressure on healthcare workers themselves, who feel obligated to compensate for the overburdened system by working harder and faster on a day-to-day basis.
This often leads to burnout, which hinders healthcare workers’ performance at work and is the primary cause of high turnover at hospitals. These conditions undercut the system’s ability to provide high-quality care, decrease the quality of life of healthcare workers, and perpetuate the cycle of a healthcare system that never reaches its full potential.
As artificial intelligence (AI) has become more advanced in recent years, technologists and healthcare professionals have noticed many opportunities for AI-based tools to improve the healthcare industry.
It can be used to help detect diseases sooner with predictive analytics, it can help triage patients more effectively in emergency settings. It can also assist doctors in finding relevant research more quickly in diagnosing patients.
But one of the most pragmatic uses of AI will be the optimization of time.
Various startups are premiering new tools that will help accomplish this goal. One example is Kahun, an Israel-based AI company, which is using automation to reduce the burden of tasks that slow down productivity in clinics and hospitals.
Kahun’s CEO, Michal Tzuchman Katz, tells us how AI can help providers spend more time with their patients, reduce healthcare worker burnout, and improve the experience of both patients and providers in interacting with the healthcare system.
Meet the Expert
Dr. Michal Tzuchman Katz, MD is the co-founder and chief medical officer of Tel Aviv-based medtech company Kahun. She began her career as a software engineer, but later decided to go back to school to pursue her dream of studying medicine.
In 2018, Dr. Tzuchman Katz partnered up with peers from her tech days to start Kahun, merging her two vocations. The company’s clinical reasoning software aims to optimize physicians’ time, focusing on the U.S. market. She continues to practice as a pediatrician in primary care and emergency settings.
The Clinical Time Crunch
Research shows that when doctors spend more time with patients, health outcomes improve and medical costs go down. But unfortunately, most doctors say they don’t spend sufficient time with patients.
According to a 2017 survey of more than 1,000 physicians by Athena Health, six out of 10 respondents agreed with statement, “My visits with patients are often too short for me to answer their questions and treat them effectively.”
This isn’t the fault of doctors. The average physician sees about 20 patients per day and works about 51 hours per week. They follow a tight schedule to make sure all the patients on their daily roster are seen. With each patient, there is a procedure that must be followed and a very short time to complete it.
The first interaction between the patient and the provider is the clinical assessment. This is when the provider will ask basic questions like, What brings you in today, what are your symptoms, and what has changed about your health in the last year? This catches the provider up on any essential health-related updates since the patient’s last visit and gives them a broad picture of their current health status.
Then, based on that information, providers may conduct physical exam(s), draw blood for tests, administer shots, or complete a number of other clinical tasks. In addition to interacting with the patient, providers must also save time to document and code new health data into the patient’s EHR.
In the end, there is little time left for providers to elaborate on their clinical reasoning or to answer any follow-up questions a patient may have before they have to move on to their next patient.
“All this needs to be performed in a very short time slot. It’d be 10 or 20 minutes, depending on the clinical setting,” says Dr. Tzuchman Katz.
“You can imagine how difficult it is to encapsulate all these tasks and do them… in such a short time and do this over and over again, patient after patient,” she says. “So the problem and the challenge is huge, yet every part of this scenario has to be done.”
The Drawbacks of EHRs
EHRs are a foundational part of a healthcare system’s operation. They store a patient’s relevant medical data such as their medical history, allergies, diagnoses, medications, treatment plans, immunization dates, radiology images, and laboratory and test results.
“The documentation is important for being able to do analysis of reports of patient data,” Dr. Tzuchman Katz says. However, it can be quite a lengthy process. “In a primary care setting, [patients with comorbidities] could have say 20 diseases in their background. [They could be prescribed] 15 medications—some of them they’re taking, some of them they’re not.”
EHRs are also typically used to complete medical coding. This is the complex process of taking a patient’s medical history, diagnoses, and treatments received, and transcribing it into universal medical codes for billing purposes.
Medical coding requires an understanding of medical and anatomical terminology and a keen attention to detail. “The doctor has to search within the EHR looking at previous blood tests that the patient had, looking at which medications that they’ve been taking, [and if] they are being compliant with their medication,” Dr. Tzuchman Katz says.
Additionally, any tests or services a patient has yet to receive must also be coded concerning the visit. “The coder will also check what the actual insurance coverage covers for that specific patient and then if the codes … are part of the coverage,” she explains. “Each patient coming into the clinic has a different coverage, so they need to be matched according to their coverage… So that makes it a very complex [process].”
If a patient’s data is coded incorrectly, it could result in a claim denial by insurance companies. Data from the Kaiser Family Foundation shows that an average 17 percent of in-network claims were denied in 2021, but insurer denial rates varied widely around this average, ranging from 2 percent to almost 50 percent.
“If an encounter was coded incorrectly, it will start a back-and-forth between the provider and the insurance company to fix these errors,” Dr. Tzuchman Katz says.
Incorrect medical coding is a major factor that prevents timely reimbursement. A survey of hospital executives found that about a third cite coding as their top concern when it comes to claim denials. The complex documentation process not only delays reimbursement and takes away from valuable doctor-patient interaction time, but it also overburdens providers.
“[Medical coding] is very important to discuss because it is one of the main reasons for burnout,” Dr. Tzuchman Katz says.
A 2020 study of 155,000 U.S. doctors found that the average physician spends about 16 minutes per patient interacting with the EHR system, which adds up to about 5.5 hours per day based on a schedule of 20 patients per day.
“Let’s say a patient who was admitted to hospital for two weeks, including an ICU admission within that stay in the hospital. The documentation of that hospital admission could be anywhere between 50 and 100 pages long,” Dr. Tzuchman Katz says.
Healthcare professionals, many of whom enter the field on the basis of wanting to help people and do meaningful work, are often fatigued by the overwhelming load of administrative work.
A nationwide survey from automation company Notable from 2022, found that 48 percent of healthcare workers are concerned about their health system’s ability to retain staff due to the repetitive tasks or documentation required in their role.
How AI Can Help
AI can improve the flow of healthcare operations and save time in a couple of key ways.
The first is by partially automating the initial clinical assessment between the provider and patient. One of Kahun’s solutions is an AI-based medical questionnaire that patients complete at home before their doctor appointments.
The interface asks patients basic questions about their health history, any new concerns, and AI-generated follow-up questions such as the severity of symptoms.
The information gathered by the AI system is automatically stored in the patient’s EHR for the doctor to review prior to the appointment so that when the actual doctor-patient interaction begins, the provider is already caught up.
“So when the patient actually comes into the room, the provider already has the [patient] history in their notes,” says Dr. Tzuchman Katz. “So they will do the cross-examination and they will ask questions to the patient, but they are not going to start from zero.”
The idea is that with the time saved during the beginning of the appointment, doctors can delve deeper with their questions, have more meaningful conversations with patients, and address patients’ follow-up questions more thoroughly.
Another way AI can help is by using predictive analytics to automate the medical coding process. “The AI can help by predicting the most relevant codes that the doctor is going to need at their very fingertips … [based on] all that patient information,” she explains.
So, rather than hunting down each individual code from the large database, the AI can make suggestions for healthcare professionals to choose from, similar to how Google lists several autocompleted suggestions as you start typing a question into its search bar.
AI-based coding tools are already doing a decent job at predicting relevant codes and in many cases, require little intervention from staff.
An article from Revcycle Intelligence reported that ER staffing company TECHealth utilized a coding AI tool that processed tens of thousands of medical charts in a matter of days. More than 80 percent of claims were handled with limited or no human interaction.
However, it’s not a flawless system. For the AI to work properly, it must be supplied with a sufficient amount of accurate data. If the healthcare provider doesn’t feed enough patient information into the system, the AI has less to work with.
“For AI to be able to help in coding, it has to understand the patient, the patient’s medical condition, and has to be able to understand the data that is within the EHR, the medical background, the medications, the risk factors, the social determinants of health—all these [pieces of] information that are within the EHR,” Dr. Tzuchman Katz says.
Additionally, if the information is not recorded in a way that the AI can understand, it could miss the mark with its generated results. Symptoms, pain levels, or other qualitative information can be difficult to measure and therefore, more difficult for AI systems to digest and configure.
It’s important to remember that these AI-based systems are just tools. They are not designed to replace humans, but to make their lives easier by automating tedious, time-consuming tasks.
Providers and professional medical coders will still need to review complex patient cases and deal with insurance companies regarding billing issues, but perhaps less often.
“The doctor of the future [will] have a much better day, have much less burnout, and much more time with the patient and practice much better medicine,” Dr. Tzuchman Katz says.
“Some of the tasks [healthcare workers] need to do are so administrative and so repetitive. So for every task that is repetitive, AI could definitely bring a lot of help and make a huge impact on how they are experiencing their work.”
Not only can AI reduce providers’ stress and burnout, it can also increase the quality of appointments and the patient’s experience.
With the time saved on administrative tasks, patients will have more room to elaborate further on their concerns and symptoms and for doctors to educate their patients.
The possibilities of the benefits of AI in healthcare are endless. As more hospitals implement AI-based tools, it will be interesting to observe if health outcomes, productivity, and burnout rates improve over time. One thing’s for sure: the future of healthcare IT is here.
