Life Style

Automated Clinical Decision Support. Is it Reliable Enough to Rely On?

Digitalization of healthcare has become irreversible and disrupted the way medical care is delivered. Health care services providers seek technologies to improve their workflow, patient experience and treatment outcomes. In our article, we’’ll elaborate on Automated Clinical Decision Support – a phenomenon intended to advise clinicians on treatment options, flag potential health issues and enhance overall medical care efficiency.

Automated Clinical Decision Support Systems, also known as CDSS or CDS, are designed to equip medical professionals and patients with targeted knowledge and patient-specific information in relevant situations to enhance medical care delivery. Technically, CDSS are sophisticated software that requires biomedical and personal data and an inferencing algorithm to present helpful recommendations for medical professionals. CDS may be implemented as a standalone app or a part of larger software solutions.

At the output, patient-tailored assessments or advice are generated and submitted to a medical professional for a decision.

Theoretically, the idea behind Automated Clinical Decision Support is to improve the clinical workflow and healthcare, inter alia:

  • Better health outcomes and personalized medical services
  • Fewer errors and adverse side effects
  • Better patient satisfaction
  • Lower information load on the clinicians.

Going from “theoretically” to “practically”, it is necessary to admit that Automated Clinical Decision Support cannot sometimes be considered as a hyper-intelligent tool to flawlessly assist clinicians. CDS is rather associated with trends than confuse medical professionals and result in worse clinical outcomes.

What is the reason for it?

Clinical decision support tools have been around for a number of years, but many of them have been somewhat standalone solutions and not well-integrated into the clinical point of care devices that people are using,” says Katherine Andriole from the U.S. Center for Clinical Data Science.

In other words, poor CDS solutions generate unnecessary alerts, reminders, distorted patient data reports and provide improper diagnostic and therapeutic support. There are several reasons and challenges for Automated Clinical Decision Support to overcome and get higher adoption rates:

  • Need to use better quality data for inferencing algorithms

CDS solutions are often integrated as a part of Electronic Health Records (EHR). The latter are not necessarily developed for further use in CDS algorithms. On the other hand, while Automated Clinical Decision Support requires accurate and comprehensive details, this information is often missing or not registered. In some cases, there can be security and privacy aspects with patient data.

In a study conducted by the U.S. National Institutes of Health Clinical Center, freely accessible and highly precise datasets were used to extract reliable data (biomarkers) from CT scans and train a learning model for further insights into patients’ medical status. Based on the results of the study, it was revealed that “automated quantitative biomarkers derived from CT scans can outperform established clinical parameters for presymptomatic risk stratification for future serious adverse events”.

  • Integration of Artificial Intelligence and machine learning

Artificial Intelligence (AI) has already become a sustainable technology trend in healthcare. As you know, medical practice proves the ability of AI-run software to diagnose breast cancer with 99 % accuracy and forecast acute kidney injury48 hours sooner than with conventional care. Keeping this in mind, Automated Clinical Decision Support could be optimized through the integration of AI and machine learning tools.

Reverting to the U.S. Center for Clinical Data Science, it has created a machine learning algorithm to identify if a patient moves during MRI. So, if a patient has had motions during the scanning, the image may lead to a distorted diagnosis. To avoid this, the smart system fueled by machine learning advises clinicians to re-scan the patient.

However, empowerment of Automated Clinical Decision Support with AI and machine learning software should be done by means of joint endeavors of technical and healthcare professionals. When CDS brings inadvertent medical results, clinicians can have a look and identify what functions to be added into the app to exclude errors, like in the above example with motions during MRI scanning.

  • Better clinicians involvement into CDS development

Further to the bullet point above, Automated Clinical Decision Support development should unite the efforts of software developers and medical professionals. This is a factor to determine if CDS can be really used for clinical purposes. Automated Clinical Decision Support can synthesize patient-related data into some evidence-based clinical guidelines at the point of medical care. However, blind acceptance of recommendations is not what we expect from clinicians. The latter can analyze and comfort them in the progress of CDS development for better clinical outcomes.

Summing it all up, we should admit that in pursuit of technologies adoption, we sometimes forget that digitalization is a team sport. Automated Clinical Decision Support still has the capacity to be a good tool to rely on. Synergy of quality data, technologies and medical knowledge will likely let CDS become the industry’s standard.

 

Back to top button