The recent World Health Innovation Forum investigated the use of artificial intelligence in the medical field. Here are some takeaways.
The World Medical Innovation Forum's recent meeting in Boston, Massachusetts, focused on artificial intelligence (AI) in health care. AI has been a hot topic lately, especially related to how it can help streamline workflow, its adoption into electronic systems to parse data, and hand-wringing over how it might upend different professions. Here are a few takeaways from the meeting:
Bringing smart machines to medicine. Smarter robotics and their entry into the health market hold promise. Using robotics for surgical purposes is a reality, but the possibility of a complete machine-guided and completed intervention is being discussed as a possibility. Other possibilities are machines for better observational power and data collection for use in the clinical decision pathway with patient monitoring technology in the hospital.
Disseminating medical expertise to areas that need it most. Communities without radiologists could upload the images and have them reviewed by software, which could then send back the results. Examples could be screening for tuberculosis and ultrasound readings, and interpretations that could provide analyses at a patient's bedside.
Getting back to face time. When we go to the doctor, an assistant takes our vitals, and then we sit awkwardly playing on our phone for the next 10 minutes, waiting for the doctor, nurse practitioner, or physician assistant to sit at a computer and ask a series of questions. They page through the electronic health records (EHRs), typing in our responses, and perhaps they use the UpToDate system for some quick decisions. Then they share their thoughts, e-scribble the script or lab order sets, and send us on our way. Maybe there was a handshake or some eye contact, but the personal touch ends there. The computer has become the focal point of medical care, and AI surprisingly is being suggested as a tool to that can help facilitate relationships between patients and practitioners.
It has been said that for every 1 hour of patient interaction, a doctor has 2 hours of documentation and administrative duties tied to it. Researchers are looking at how administrative duties can be automated, reducing the computer workload for medical professionals. One interesting highlighted example was an AI program that would serve as a co-pilot for providers to help analyze medical information, make recommendations, and take care of billing during a patient visit.
Harnessing the power of digital pathology. Way outside of the realm of pharmacy but interesting nonetheless is the use of AI to look at pathological specimens and analyze them.
Improving health with digital devices. Mobile health tools can aid in data tracking and nudging patients toward healthy behavior patterns. Nonetheless, we are still in the infancy of the technology. The issue, ultimately, is which data are meaningful to use and beneficial for clinical practice. The other question is how to then trawl that information and use if for predictive health outcomes and intervene appropriately. Examples are the use of continuous blood pressure and blood glucose monitoring that pharmacists could integrate into medication therapy management and for clinical duties.
Melding mind and machine. Essentially, how can we tap into what the brain is doing with current technology and figure out how to take the electrochemical framework and reproduce it into actionable events? This is more than, say, thinking of a message and having it then sent via phone as a text message to a friend. Instead, a more realistic and closer possibility would be to determine what processes are involved in movement. This could have an excellent application for prosthetics and other patients with neurological diseases or disabilities who could benefit from such technology.
Minimizing the threats of antimicrobial resistance and infections associated with antibiotic use. It is not an exaggeration to say that the danger posed by the inappropriate use of antibiotics during this past century could put the species at risk for a post-antibiotic era, where people will die again at high rates of infections not seen since the 19th century. However, AI shows some promise in being used to help combat the danger of infectious diseases, to stratify the risk of patients for infections. For instance, it may be possible to identify patients at high risk of infection with clostridium difficile and determine when proactive approaches are needed alongside treatment. Other nosocomial infections could also be classified, as could the length of stay, the reason for hospitalization, rates of infections, etc. This approach, at least, would serve as a possible safety mechanism to identify high-risk patients and possibly scale back the rates of infections and reduce the spread of disease into the community at large.
Reading the tea leaves of cancer immunotherapy. The research into cancer immunotherapies has garnered much attention, but not all patients have seen success. After all, not every patient has an immune system that can fight cancer. Identifying those biomarkers to determine which patients may benefit from immunotherapy is still being investigated, and the use of some technology to help with this is critical.
Tapping next-gen radiology. Radiological images for years have been on screens and analyzed by radiology specialists. Yet, the question has become whether these high-cost specialists are worth the cost. Surely, if it is just images, computer software can review them quickly and perhaps even more accurately. Researchers are looking into whether such AI programs can do prescreenings or serve as adjunctive tools for clinicians in their workflow. There have been rumblings that this could rock the radiology profession, but the reality is that this type of technology would not replace radiologists but assist them.
Using EHRs to predict disease risk. Precision medicine and figuring out what treatment works for each patient is the end game right now. But moving from a population treatment model to precision medicine requires a lot of work. Arguably, the benefit of EHR systems being put into place is the aggregation of large pools of datasets that can be analyzed. If we can look at the data and see when patients progress from one stage of a disease to another, we might be able to come up with algorithms that can predict this in future patients and hopefully, provide better care.
Disruptive Dozen. 2018 World Medical Innovation Forum. worldmedicalinnovation.org/wp-content/uploads/2018/04/Partners-FORUM-2018-BROCHURE-D12-AI-180410_1202-FREV2-FOR-WEB-X3.pdf. Accessed May 7, 2018.