Clinical Decision Support Systems
A brief overview of clinical decision support systems, how they're currently implemented, and ways in which they must be improved.
Oliver |
Creating algorithms and tools that support pathologists and radiologists to diagnose patients’ tumor status and metastasis is the main reason I apply for BHI in UW, which I believe I only understand part of the goal, use case, and definition of CDSS at that time. With the readings and assignments that focus on EHRs and CDSS in these weeks, I have a better understanding about what EHR is, how EHR relates to CDSS, and the challenges we are facing. If we take the definition of CDSS as the approach to improve healthcare delivery by assisting the decision making process with clinical knowledge, and massive data, it’s not hard to realize that the focus is in decision making but not any fancy algorithms or techniques. I really appreciate that Dave mentions the AMIA competency framework in our first class which I think it successfully guides me to think about what we can do as an informatician to solve the problem. I personally am still more interested in the computer science aspect that uses machine learning and NLP in CDSS to support physicians to diagnose; however, I’ll keep in mind that the successful design and implementation of EHRs and CDSS should not only focus on those techniques, but need to take into consideration about human factors and other data interoperability issues that current CDSS face.
|
Weipeng |
One thought I have on this clinical decision support system chapter is that I feel that the current state of the art still has a lot of rooms to improve but I am optimistic. It might not be the best time for doing research on the clinical decision support system but it is also not the worst time. With the advance of artificial intelligence and the amount of data gathered becomes larger, there are more possibilities. I am also glad that we discovered the usability of the clinical decision support system to be an important factor and more researches is done is this field. I personally think this is the correct direction. In order to help doctors, we should be able to first understand them and the problem they are facing. There was also exciting news about the use of a clinical decision support system to fight against COVID-19. In the last month or so when Wuhuan was at its more dire situation, systems from Tencent, an IT company, were sent to Wuhan to assist physicians checking thousands of lung screens. The image recognization AI was able to tell the possibility of COVID-19 from lung screens. I think this is a good example of CDSS implementation.
|
Dakota |
It is disappointing to learn more about the ways in which digital health systems are, generally, not performing at the level we wish they were. Out of all the pieces of health IT information that I'm aware of, I think that CDSS has the greatest potential for massive improvements in the healthcare system so it is unfortunate, yet not surprising, to know that, as it currently is, CDSS is not only not very helpful, but in many cases, doing destructive work in healthcare environments. Here are a few things that I found interesting when doing the research to write this article:
Many CDSS are strictly the artificial intelligence that is used by the system and not necessarily the system itself. As mentioned in our writing above, CPOE is sometimes a "secondary" aspect to the CDSS which feels bizarre and incredibly inefficient to me. It seems odd that in order for a company to sell a CDSS to a provider organization they will often develop the CPOE themselves and then license the CDSS software from a third party company in order to integrate the provider-computer aspect of care with the actual artificial intelligence which performs the data analyses. I suspect that perhaps part of the problems experienced with CDSS (particularly the user-interface issues) are caused by the fact that software developers think of CDSS and CPOE as two separate softwares rather than two facets of code that work in tandem to accomplish a single goal. In writing this other articles, we have used a webpage set up by a division of the US government (healthit.gov) that is dedicated to the documentation of health IT systems. In reading through the health IT article on CDSS, it suddenly became very apparent to me that this website is a heavily biased information source. In regards to many of the health IT systems that we write about in these modules, this government website presents only the positive aspects of the various technologies and further presents ideal systems as truths. For example, the webpage on CDSS says explicitly that " person-specific information, intelligently filtered or presented at appropriate times, to enhance health and health care," which, as we've just spent this entire module discussing, is simply not true. Yes, this is the ideal situation, but we are seeing that this is rarely the case. It is unsurprising that they would present information in such a highly-skewed way as they are pushing very hard for US-healthcare to adopt these technologies and would only want to sell them on the positive aspects, however, it is concerning to me if for no other reason that usually, when you google "clinical decision support" that webpage is the first search result which is simply full of extremely incomplete information. |