Trends

AI and Clinical Decision Support in Healthcare

Clinical decision support (CDS) system is an application which analyzes patient data with available health information and helps providers make right clinical decisions through out the clinical work flow

CDS helps health care professionals make decisions in various stages such as preventive care, diagnosis, treatment etc. by provided relevant clinical information, alerts or drug formulary guidelines.

In this article, we will see how CDS application supports healthcare, its barriers and success factors. We will also identify how to improve the clinical decision work flow and adopt the best approach in decision making.

How can CDS support Healthcare?

Clinical decision support system assimilates computerized knowledge based clinical information that prompts the provider with the relevant recommendations for the patient to improve care. Using the information as suggestion the provider diagnoses and analyzes medical cases. CDS’s main aim is to identify diseases at an earlier stage by analyzing various type of tumor tissues, patient related clinical data and different characteristics of diseases along with symptoms. A good CDS should be able to recommend drug dosage alerts, reminders and suggestions, aid in CPOE, diagnosis, therapies etc.

Risk Factors:

Lack of knowledge base: CDS works mainly with evidence obtained from the information available and patient  details. If the system is not updated with the latest disease patterns and diagnosis, there is a chance of using outdated information for decision making, which causes inefficiency in treatment plan.

Less User friendly: Recommendations/ guidelines are sometimes lengthy, medical image details might not be clear and may not be easy to access. In such cases, there might be a possibility of wrong diagnosis which can lead to delay in medical treatments. When there is no proper suggestion from the system due to lack of symptoms related information, users need to dedicate extra time to perform additional disease analysis themselves.

Conflicts:  When there is a conflict between the users thought process and clinical information system in decision making, users might be disappointed with the recommendation. The conflict could arise because of several reasons including the limited scope of the system or the gut feel of the doctor.

Patient Specific Diagnosis: Some of the disease diagnosis might entirely depend on genetic, age, gender, race and cannot depend on previous diagnosis details.

Frequent data update: As disease symptoms and diagnosis are changing over the generation, system needs frequent data updating to provide right information.

How to overcome?

CDS system should allow the user to continuously feed evidence related details like updated drug information, symptoms and other type of details to avoid out dated information.

During the decision making process, CDS can give options to accept or reject the recommendations/ plan of care suggested by the system and it can take the reason from the healthcare users for rejecting specific suggestions. When certain suggestions are rejected repeatedly, then that respective area would be identified for improvement.

When system can provide genetics risk information and family history details in depth, appropriate treatments and diagnosis can be given for patients.

Options should be given to users to choose the areas to be enabled with CDS services like alerts, reminders, suggestions and guidelines in the healthcare systems.

When CDS suggestions are modified by users, it needs to be recorded in the database. At an organizational level, these changes need to be tracked to get revised plans. Latest scientific and medical knowledge to be updated in clinical settings.

CDS can be augmented with Artificial Intelligence to get better suggestions and treatment plans. As AI can process data from various sources like internet, voice, care providers etc., CDS can use this data to provide better care and services.

How AI can support CDS?

AI algorithms can perform functions such as symptoms validation, treatment suggestion, tissue recognition and image detection. AI in CDS can be used by extracting data from various sources and comparing with patient symptoms to find possible diseases.

Below are few areas where CDS can be used more effectively by using AI:

Diagnosis detection: Designing machine learning (Neural Networks) devices with CDS can detect infections and conditions. Data related to blood, tissues, biopsy results can be used and compared with specific patient results and appropriate treatment plan can be made.

Big data: Large data collections can be managed with big data analytics and thus data can be used by machines to efficiently learn diseases using trial and error methods.

Precision medicine and genomics:  By identifying large sets of genomic data a, the system would have an ability to link diseases with DNA. AI algorithms can  help to find the abnormal tumor, tissues by comparing with the biological details.

Medical Imaging: AI can help with imaging analysis and examine abnormal cases and directly provide intelligence into EMR.

Drug Discovery: AI algorithms can compare and find molecules from paramedical data and develop new drugs.

Best approach:

A combination of a knowledge based methodology and AI technology would complement each other and would be a great fit to implement effective CDS system.

Knowledge based methodology can speed up the system if used with simplified guidelines, and a flexible knowledge repository would be essential for conducting evidence based medicine.

AI is a developing technology and adopting AI in CDS would result in better a CDS system, which would evolve with the technology. AI could bring in a much higher level of intelligence that CDS can access, and also couple with technologies like big data analytics and deep learning.

Conclusion:

To adapt to Healthcare changes in future with the advent of value based care and evidence based medicine,  AI can be used with CDS to increase accuracy and performance in Healthcare systems. Combining knowledge based methodology with AI technology would be the best approach to get the possible outcomes. With data playing a major role in the coming years, in the near future doctors might issue data intelligence reports generated by CDS systems along with their medical reports to patients and back their health reports with sufficient evidence.

References:

  1. https://pdfs.semanticscholar.org/aa26/f01dd74604d7f5066dd0826beb1ac2563de3.pdf
  2. Improving Outcomes with Clinical Decision Support: An Implementer’s Guide by By Jerome A. Osheroff, MD, FACP, FACMI, Jonathan M. T
  3. https://www.bmj.com/content/316/7149/1959.full
  4. https://www.bmj.com/content/330/7494/765.short
  5. https://www.dicardiology.com/article/advances-clinical-decision-support-software
  6. https://www.researchgate.net/publication/305915394_Artificial_Intelligence_in_Clinical_Decision-Making_Will_It_Cure_the_Field_of_Medicine
  7. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4381462/
  8. https://www.cerner.com/ae/en/blog/clinical-decision-support-systems-in-the-era-of-ai

   Authored by:

     Mohanapriya Kolanjinadhan

      Sr. Software Tester

     http://incarnus.com/

Tagged ,

About Incarnus

We are a global provider of next-generation, cloud-enabled solutions for all levels of healthcare from primary health practice to large-scale hospital networks.
View all posts by Incarnus →

1 thought on “AI and Clinical Decision Support in Healthcare

  1. Thanks for the valuable post! AI can help make medical equipment smarter, imaging results faster, and examinations more precise. AI will enable a world of data where one patient’s history is transformed through insights from the cohort and where patient populations are enriched by the data of each individual.

Leave a Reply

Your email address will not be published. Required fields are marked *