2024 FSA Podium and Poster Abstracts

S014: AUTOMATING DELIRIUM DETECTION: A NOVEL DIAGNOSTIC DEVICE IN THE PERIOPERATIVE SETTING
Camila Teixeira; Simone Phang-Lyn; Eva Chen, MD; Ashton Huey; Jeff Jacobs, MD, MBA; Steven Minear, MD, MBA, FASA; Cleveland Clinic Florida

Delirium is a disturbance in attention and awareness characterized by an acute onset and fluctuating course1. Delirium is a post-surgical complication, relatively common in patients older than 65 years, and usually arises in the first three postoperative days. Evaluating a patient’s mental status with enough frequency to facilitate early intervention is logistically difficult and is worsened by persistent hospital staffing challenges.  To address the challenges of delirium assessment, our group piloted a novel device to automate delirium detection. We present a proof-of-concept device that alleviates the burden of delirium data capture and identifies patients in need of intervention.  Patients 60 years or older undergoing surgical procedures requiring admission to the Surgical Intensive Care Unit (SICU) were included. Beginning the first extubated postoperative day, subjects responded to both automated and manual CAM-ICU assessments, twice a day, for 3 days or until ICU discharge. Patients also underwent baseline testing. Sensitivity, specificity, and accuracy were calculated to compare assessments.

50 patients undergoing cardiac surgery were screened and 19 were enrolled. CAM-ICU results from both assessments (automated and manual) were compared. Our delirium assessment monitor was positioned in enrolled patient’s rooms, functioning as a touch-screen tablet with computer vision (Figure 1). 15 patients had paired-wise test result values available for analysis. For these patients, the mean ± SD of age was 69.6 ± 6.3 with 3 (20%) females and 12 (80%) males. There was a total of 69 paired-wise test result values (Table 1). The sensitivity of the automated monitor was 21% and the specificity was 98%, with an accuracy of 82% (Table 2).

To our knowledge, this is the first use case of an automated delirium data-capture and assessment tool. This tool is particularly useful in the ICU, where staff have time-sensitive obligations towards patient stabilization. A device capable of gathering delirium data with high reliability can alert the care team of delirium. A study limitation is that the sensitivity is relatively low, which may have resulted from either difficulty interacting with the monitor in an ICU environment or a low sample size. The device is currently undergoing revision, particularly to overcome the noise issues common in an ICU environment.

In the next phase, our group plans to incorporate patient monitor signals into the delirium algorithm, capturing hemodynamic and laboratory values in the hours or days prior to delirium signals. We plan to take a machine-learning approach to develop this tool towards both to diagnosing and predicting delirium. With prompt, accurate diagnoses, better treatments can be developed.

1.     Inouye, S. K. et al. Clarifying confusion: the confusion assessment method. A new method for detection of delirium. Ann. Intern. Med. 113, 941–948 (1990).

2.     Franco, K., Litaker, D., Locala, J. & Bronson, D. The cost of delirium in the surgical patient. Psychosomatics 42, 68–73 (2001).

3.     Guenther, U. et al. Validity and reliability of the CAM-ICU Flowsheet to diagnose delirium in surgical ICU patients. J. Crit. Care 25, 144–151 (2010).