he prospect of improved clinical outcomes and more efficient health systems has fueled a rapid rise in the development and evaluation of AI systems over the last decade. Because most AI systems within healthcare are complex interventions designed as clinical decision support systems, rather than autonomous agents, the interactions among the AI systems, their users and the implementation environments are defining components of the AI interventions' overall potential effectiveness. Therefore, bringing AI systems from mathematical performance to clinical utility needs an adapted, stepwise implementation and evaluation pathway, addressing the complexity of this collaboration between two independent forms of intelligence, beyond measures of effectiveness alone 1 . Despite indications that some AI-based algorithms now match the accuracy of human experts within preclinical in silico studies 2 , there
Background
Artificial intelligence–based clinical decision support (AI-CDS) tools have great potential to benefit intensive care unit (ICU) patients and physicians. There is a gap between the development and implementation of these tools.
Objective
We aimed to investigate physicians’ perspectives and their current decision-making behavior before implementing a discharge AI-CDS tool for predicting readmission and mortality risk after ICU discharge.
Methods
We conducted a survey of physicians involved in decision-making on discharge of patients at two Dutch academic ICUs between July and November 2021. Questions were divided into four domains: (1) physicians’ current decision-making behavior with respect to discharging ICU patients, (2) perspectives on the use of AI-CDS tools in general, (3) willingness to incorporate a discharge AI-CDS tool into daily clinical practice, and (4) preferences for using a discharge AI-CDS tool in daily workflows.
Results
Most of the 64 respondents (of 93 contacted, 69%) were familiar with AI (62/64, 97%) and had positive expectations of AI, with 55 of 64 (86%) believing that AI could support them in their work as a physician. The respondents disagreed on whether the decision to discharge a patient was complex (23/64, 36% agreed and 22/64, 34% disagreed); nonetheless, most (59/64, 92%) agreed that a discharge AI-CDS tool could be of value. Significant differences were observed between physicians from the 2 academic sites, which may be related to different levels of involvement in the development of the discharge AI-CDS tool.
Conclusions
ICU physicians showed a favorable attitude toward the integration of AI-CDS tools into the ICU setting in general, and in particular toward a tool to predict a patient’s risk of readmission and mortality within 7 days after discharge. The findings of this questionnaire will be used to improve the implementation process and training of end users.
In the version of this article initially published, a list of the DECIDE-AI expert group members and their affiliations was omitted and has now been included in the HTML and PDF versions of the article.
Identification of postoperative infections based on retrospective patient data is currently done using manual chart review. We used a validated, automated labelling method based on registrations and treatments to develop a high-quality prediction model (AUC 0.81) for postoperative infections.
Background
C-reactive protein (CRP) is a nonspecific marker of inflammation and due to surgery alone CRP levels are increased post-operatively. Although multiple studies investigated the potential of post-operative CRP levels to predict post-operative infection, its discriminative capacity remained unclear. We aimed to describe the kinetics of CRP for post-operative patients with- and without a post-operative infection for different types of surgery in the first seven post-operative days.
Methods
We conducted a single center cohort study in adult patients undergoing surgery between 2011 and 2021 with the use of whole digital case file extraction (big data). All patients that had either an infectious complication registered, non-prophylactic antibiotics prescribed and/or an infection related surgical re-intervention within 7 days after having undergone surgery, were labelled as having an infection. We analyzed the CRP kinetics in the first seven post-operative days for all these patients.
Results
A total of 39.985 individual patients were included of which 12.002 patients had at least one CRP measurement and 4.014 patients had an infection in the first seven post-operative days. CRP was measured in 66% of the patients with an infection and in 26% of the patients without an infection. In all patients CRP increased during the first two post-operative days regardless of infection status (Figure 1). In cardiothoracic, orthopedic, ear-nose-throat, general and urological surgery CRP decreased after day 2 or 3 in all patients but the absolute CRP value remained higher in the following days in patients with an infection compared to patients without an infection. In gynecological and oral surgery CRP decreased on day 3 in patients without an infection but not before day 4 in patients with an infection. CRP was not discriminative in neurosurgical or plastic surgery patients (Figure 2).
CRP kinetics in the first seven post-operative days for patients with- and without an infection
CRP kinetics for different subspecialties in the first seven post-operative days
CTS = cardiothoracic surgery; ENT = ear-nose-throat surgery; GEN = general surgery; GYN = gynecological surgery; NEUR = neurosurgery; ORAL = oral surgery; ORT = orthopedic surgery; PLA = plastic surgery; URO = urological surgery.
Conclusion
In the first 2 post-operative days CRP increases in both patients with- and without an infection which is probably due to the surgical procedure itself. After the second post-operative day differences in CRP kinetics develop between patients with- and without an infection, but how they differ depends, among other factors, on the type of surgery.
Disclosures
Siri L. van der meijden, MSc, Healthplus.ai: Employee Bart F. Geerts, MD. PhD., Healthplus.ai: Board Member.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.