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... Show moreBackground: 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. Show less
Lindow, T.; Pahlm, O.; Olson, C.W.; Khoshnood, A.; Ekelund, U.; Carlsson, M.; ... ; Engblom, H. 2020
Electrocardiographic Decision Support - Myocardial Ischaemia (EDS-MI) is a graphical decision support for detection and localization of acute transmural ischaemia. A recent study indicated that EDS... Show moreElectrocardiographic Decision Support - Myocardial Ischaemia (EDS-MI) is a graphical decision support for detection and localization of acute transmural ischaemia. A recent study indicated that EDS-MI performs well for detection of acute transmural ischaemia. However, its performance has not been tested in patients with non-ischaemic ST-deviation. We aimed to optimize the diagnostic accuracy of EDS-MI in patients with verified acute coronary occlusion as well as patients with non-ischaemic ST deviation and compare its performance with STEMI criteria. We studied 135 patients with non-ischaemic ST deviation (perimyocarditis, left ventricular hypertrophy, takotsubo cardiomyopathy and early repolarization) and 117 patients with acute coronary occlusion. In 63 ischaemic patients, the extent and location of the ischaemic area (myocardium at risk) was assessed by both cardiovascular magnetic resonance imaging and EDS-MI. Sensitivity and specificity of ST elevation myocardial infarction criteria were 85% (95% confidence interval (CI) 77, 90) and 44% (95% CI 36, 53) respectively. Using EDS-MI, sensitivity and specificity increased to 92% (95% CI 85, 95) and 81% (95% CI 74, 87) respectively (p=0.035 and p<0.001). Agreement was strong (83%) between cardiovascular magnetic resonance imaging and EDS-MI in localization of ischaemia. Mean myocardium at risk was 32% (+/- 10) by cardiovascular magnetic resonance imaging and 33% (+/- 11) by EDS-MI when the estimated infarcted area according to Selvester QRS scoring was included in myocardium at risk estimation. In conclusion, EDS-MI increases diagnostic accuracy and may serve as an automatic decision support in the early management of patients with suspected acute coronary syndrome. The added clinical benefit in a non-selected clinical chest pain population needs to be assessed. Show less
Background Control of blood glucose levels is needed not only to alleviate symptoms of hypoglycaemia and hyperglycaemia, but also to prevent or delay diabetes-related complications. Advice for... Show moreBackground Control of blood glucose levels is needed not only to alleviate symptoms of hypoglycaemia and hyperglycaemia, but also to prevent or delay diabetes-related complications. Advice for glucose control is usually provided to patients by members of the health care team. However, many diabetes apps claim to enhance self-management of blood glucose by providing decision support to patients when an out-of-range blood glucose level is recorded. In this study, we investigated the appropriateness of action prompts provided by diabetes apps for hypoglycaemia and hyperglycaemia against evidence-based guidelines. Methods We used methods previously reported to identify and select diabetes apps, which were downloaded and assessed against the American Diabetes Association (ADA) guidelines. Screenshots of action prompts corresponding to low or high out-of-range blood glucose values were subjected to content analysis. Results Of 371 diabetes self-management apps evaluated, only 217 and 216 apps alerted patients about hypoglycaemia and hyperglycaemia, respectively. Of these, 20.7% (45/217) and 15.3% (33/216) also provided action prompts. We found 5.1% of apps (hypoglycaemia: 11/217; hyperglycaemia: 11/216) provided prompts that were either too general to be helpful or not aligned with ADA guidelines. Overall, only 17.9% (39/217) and 14.8% (32/216) provided appropriate action prompts for hypoglycaemia and hyperglycaemia, respectively. Conclusion Less than one fifth of apps provided evidence-based steps to guide patients through hypoglycaemia and hyperglycaemia. The majority of apps failed to provide just-in-time diabetes self-management education to prevent frequent or severe episodes of hypoglycaemia and hyperglycaemia. Our findings emphasize the need for better design and quality assurance of diabetes apps. Show less
Schneider, A.J.T.; Besselink, P.L.; Zonderland, M.E.; Boucherie, R.J.; Hout, W.B. van den; Kievit, J.; ... ; Rabelink, T.J. 2018
