Is artificial intelligence (AI) the panacea to cure healthcare’s biggest challenges? Will new AI applications and advanced analytics tools provide clinicians with the point-of-care information they need to optimize patient outcomes?
If you are a clinician or a healthcare CEO, you might very well believe that AI technologies are on track to transform healthcare. After all, who hasn’t read the headlines about ground-breaking solutions that are automatically analyzing volumes of clinical data and offering precise diagnoses and personalized therapies that drive more effective and efficient patient care?
If you are a healthcare CIO, you’ve probably read those same headlines, yet are saying to yourself, “Hold on people – not so fast.”
It’s not that CIOs aren’t optimistic about the potential of AI. I know quite a few CIOs who can get downright geeky theorizing about all the possible AI healthcare use cases. It’s exciting and that cannot be ignored. But, not surprisingly, these same CIOs, are simultaneously skeptical about the ability of our current IT systems to deliver the top-quality, accurate and usable data that AI solutions require to perform their algorithmic magic.
Fueling the AI engine to produce insights
Healthcare CIOs understand that without ready-access to quality data to feed AI engines, the technology is simply not going to work. Traditional EHRs, which were originally designed to support billing, typically store the bulk of their information as unstructured and uncorrelated data, including free-text within chart notes and in pathology and radiology reports. Rarely do these EHRs maintain precise data points on highly relevant clinical details such as patient presentation or particulars related to social determinants of health.
A Deep-dive Into Specialty Pharma
A specialty drug is a class of prescription medications used to treat complex, chronic or rare medical conditions. Although this classification was originally intended to define the treatment of rare, also termed “orphan” diseases, affecting fewer than 200,000 people in the US, more recently, specialty drugs have emerged as the cornerstone of treatment for chronic and complex diseases such as cancer, autoimmune conditions, diabetes, hepatitis C, and HIV/AIDS.
As much as 80 percent of the patient information within EHRs is stored as free-text and not mapped to data standards. CIOs realize that in order to draw quality conclusions from AI queries, you first need data that’s in a codified and structured format. CIOs are cognizant of the fact that we won’t realize the full potential of AI for healthcare until we clean up and organize the vast amounts of clinical data created over the last couple of decades as EHR adoption has grown.
Contrary to what marketeers may want us to believe, we are still in the very early days of AI and machine-learning for healthcare. Mainstream AI applications are not yet sophisticated enough to sift through all the unstructured and uncorrelated pieces of data in an EHR and generate the clean, detailed data required for AI insights.
For example, providers would likely love solutions that leverage natural language processing (NLP) and have the ability to convert dictated chart notes to free text, and then free text to data that’s stored in an actionable format. Unfortunately, the error rates for converting speech to text to data are, at best, between 8 and 10 percent – which is unacceptably high for applications that require the data for safe and effective clinical decision-making.
The reality of AI is that it will fail to fix healthcare’s big challenges until organizations have the technology to link disparate clinical data and concepts and map them to standard nomenclatures such as ICD-10, SNOMED, RxNorm, and LOINC. Users could then easily filter duplicate information and data could be organized into a structured format that provides AI technologies with the solid starting point required to make a positive impact on healthcare.
Remembering clinicians—and their patients
CIOs also know never to mess with clinician productivity. In time we will see more tools to help enterprises organize their vast volumes of clinical data, which in turn will drive wider adoption of AI solutions in healthcare. As organizations evaluate new solutions, stakeholders must be wary of any alternatives that compromise clinician productivity or even more troublingly, compromise patient care.
Clinicians won’t embrace AI technologies that interfere with existing workflows and diminish physician productivity or that put the knowledge of the machine above their own knowledge and experience. Thus, any filtering and organizing of data must be performed behind the scenes to optimize the usability of EHRs, and never stand in the way of physicians’ clinical decision making at the point-of-care.
CIOs recognize that AI solutions are not the panacea to cure healthcare – at least not yet. However, as providers leverage technologies to organize and clean up their data, they’ll in time have more structured and query-friendly information to feed AI engines and create more valuable – and transformative – insights.
Photo: ANDRZEJ WOJCICKI, Getty Images
