Deciding Thrombolysis in AIS Based on Automated versus on WhatsApp Interpreted ASPECTS, a Reliability and Cost-Effectiveness Analysis in Developing System of Care

Abstract

Background: Automated ASPECTS has the potential of reducing interobserver variability in the determination of early ischemic changes. We aimed to assess the performance of an automated ASPECTS vs. ASPECTS interpreted for sent CT images on WhatsApp and to correlate these results with the outcome.Materials and Methods: Patients with anterior circulation stroke who had baseline NCCT and underwent successful IV-thrombolysis were included. NCCT-ASPECTS was assessed by two neuroradiologists, and discrepancies were resolved by agreement. Two groups of patients were included; group 1, where treatment was decided after an automated ASPECTS interpretation that was provided by RAPID software, and group 2, where patients received IV-tPA after an assessment of CT images sent on WhatsApp.Results: A total of 122 patients were included: 36 in group 1 and 86 in group 2. In group 2, the interobserver agreement for NCCT ASPECTS was moderate (κ = 0.36), as was the dichotomized data (κ = 0.44). IOA, however, improved (to κ = 0.57 and κ = 0.64) when the same CT images were interpreted on a workstation. In group 1, Automated ASPECTS showed excellent agreement (κ = 0.80) with agreement reads for workstation images. There were significantly (P < 0.001) increased odds of functional independence and fewer hemorrhagic complications with thrombolyzed patients in group 1.Conclusions: Automated ASPECTS provided by the RAPID@IschemaView ASPECTS performs at a level equal to the agreement read of expert neuroradiologists, and this performance was severely degraded when WhatsApp captured CT images used for ASPECTS assessment. In our study, we found that automated ASPECTS might predict outcomes after IV thrombolysis.

Systematic Review of Health Economic Evaluations Focused on Artificial Intelligence in Healthcare: The Tortoise and the Cheetah

“…The remaining 20 studies were read full text. [19][20][21][22][23][24][25][26][27][28][29][30][31][32][33][34][35][36][37][38] The relevant data were extracted and the 2 reviewers independently further assessed the quality of the articles per the CHEERS and Philips checklists.…”

“…A general overview of the included studies, including the details of the reported AI applications, is provided in Table 2. [19][20][21][22][23][24][25][26][27][28][29][30][31][32][33][34][35][36][37][38] The majority of studies was published in 2019 or later. The studies were conducted in a range of medical specialties, yet ophthalmology was evidently the dominant field.…”

“…20,21,28,29 The most common type of AI was automated medical image analysis models, as 9 studies (45%) evaluated this type of AI. [19][20][21][22][27][28][29][30]32 A total of 6 studies (30%) evaluated pattern recognition in clinical data for predictive modeling. [24][25][26][36][37][38] Nevertheless, all phases of the care pathway from prevention to follow-up were supported by AI in at least 1 included study; screening and treatment monitoring (intervention) were the most prevalent phases of the care pathway in which AI was applied.…”

“…23 A complete overview of the health economic methodological details of the included studies can be found in Table 3. [19][20][21][22][23][24][25][26][27][28][29][30][31][32][33][34][35][36][37][38]…”

Validation of automated Alberta Stroke Program Early CT Score (ASPECTS) software for detection of early ischemic changes on non-contrast brain CT scans

“…A similar study analyzing relative Hounsfield unit density per region found a sensitivity of 45% and a specificity of 93% for a HU ratio threshold of < 0.94 and an area under the curve of 0.780 [ 24 ]. A study assessing RAPID@IschemaView automated ASPECTS found a performance equal to the agreement read of expert neuroradiologists [ 25 ].…”