Machine Learning Prediction Models to Reduce Length of Stay at Ambulatory Surgery Centers Through Case Resequencing.

The post-anesthesia care unit (PACU) length of stay is an important perioperative efficiency metric. The aim of this study was to develop machine learning models to predict ambulatory surgery patients at risk for prolonged PACU length of stay - using only pre-operatively identified factors - and then to simulate the effectiveness in reducing the need for after-hours PACU staffing. Several machine learning classifier models were built to predict prolonged PACU length of stay (defined as PACU stay ≥ 3 hours) on a training set. A case resequencing exercise was then performed on the test set, in which historic cases were re-sequenced based on the predicted risk for prolonged PACU length of stay. The frequency of patients remaining in the PACU after-hours (≥ 7:00 pm) were compared between the simulated operating days versus actual operating room days. There were 10,928 ambulatory surgical patients included in the analysis, of which 580 (5.31%) had a PACU length of stay ≥ 3 hours. XGBoost with SMOTE performed the best (AUC = 0.712). The case resequencing exercise utilizing the XGBoost model resulted in an over three-fold improvement in the number of days in which patients would be in the PACU past 7pm as compared with historic performance (41% versus 12%, P<0.0001). Predictive models using preoperative patient characteristics may allow for optimized case sequencing, which may mitigate the effects of prolonged PACU lengths of stay on after-hours staffing utilization.

Identification of Preanesthetic History Elements by a Natural Language Processing Engine.

BACKGROUND: Methods that can automate, support, and streamline the preanesthesia evaluation process may improve resource utilization and efficiency. Natural language processing (NLP) involves the extraction of relevant information from unstructured text data. We describe the utilization of a clinical NLP pipeline intended to identify elements relevant to preoperative medical history by analyzing clinical notes. We hypothesize that the NLP pipeline would identify a significant portion of pertinent history captured by a perioperative provider. METHODS: For each patient, we collected all pertinent notes from the institution's electronic medical record that were available no later than 1 day before their preoperative anesthesia clinic appointment. Pertinent notes included free-text notes consisting of history and physical, consultation, outpatient, inpatient progress, and previous preanesthetic evaluation notes. The free-text notes were processed by a Named Entity Recognition pipeline, an NLP machine learning model trained to recognize and label spans of text that corresponded to medical concepts. These medical concepts were then mapped to a list of medical conditions that were of interest for a preanesthesia evaluation. For each condition, we calculated the percentage of time across all patients in which (1) the NLP pipeline and the anesthesiologist both captured the condition; (2) the NLP pipeline captured the condition but the anesthesiologist did not; and (3) the NLP pipeline did not capture the condition but the anesthesiologist did. RESULTS: A total of 93 patients were included in the NLP pipeline input. Free-text notes were extracted from the electronic medical record of these patients for a total of 9765 notes. The NLP pipeline and anesthesiologist agreed in 81.24% of instances on the presence or absence of a specific condition. The NLP pipeline identified information that was not noted by the anesthesiologist in 16.57% of instances and did not identify a condition that was noted by the anesthesiologist's review in 2.19% of instances. CONCLUSIONS: In this proof-of-concept study, we demonstrated that utilization of NLP produced an output that identified medical conditions relevant to preanesthetic evaluation from unstructured free-text input. Automation of risk stratification tools may provide clinical decision support or recommend additional preoperative testing or evaluation. Future studies are needed to integrate these tools into clinical workflows and validate its efficacy.

Clinical Cybersecurity Training Through Novel High-Fidelity Simulations.

BACKGROUND: Cybersecurity risks in health care systems have traditionally been measured in data breaches of protected health information, but compromised medical devices and critical medical infrastructure present risks of disruptions to patient care. The ubiquitous prevalence of connected medical devices and systems may be associated with an increase in these risks. OBJECTIVE: This article details the development and execution of three novel high-fidelity clinical simulations designed to teach clinicians to recognize, treat, and prevent patient harm from vulnerable medical devices. METHODS: Clinical simulations were developed that incorporated patient-care scenarios featuring hacked medical devices based on previously researched security vulnerabilities. RESULTS: Clinicians did not recognize the etiology of simulated patient pathology as being the result of a compromised device. CONCLUSIONS: Simulation can be a useful tool in educating clinicians in this new, critically important patient-safety space.

Ultrasound-guided central venous access using Google Glass.

BACKGROUND: The use of ultrasound during invasive bedside procedures is quickly becoming the standard of care. Ultrasound machine placement during procedures often requires the practitioner to turn their head during the procedure to view the screen. Such turning has been implicated in unintentional hand movements in novices. Google Glass is a head-mounted computer with a specialized screen capable of projecting images and video into the view of the wearer. Such technology may help decrease unintentional hand movements. OBJECTIVE: Our aim was to evaluate whether or not medical practitioners at various levels of training could use Google Glass to perform an ultrasound-guided procedure, and to explore potential advantages of this technology. METHODS: Forty participants of varying training levels were randomized into two groups. One group used Google Glass to perform an ultrasound-guided central line. The other group used traditional ultrasound during the procedure. Video recordings of eye and hand movements were analyzed. RESULTS: All participants from both groups were able to complete the procedure without difficulty. Google Glass wearers took longer to perform the procedure at all training levels (medical student year 1 [MS1]: 193 s vs. 77 s, p > 0.5; MS4: 197s vs. 91s, p ≤ 0.05; postgraduate year 1 [PGY1]: 288s vs. 125 s, p > 0.5; PGY3: 151 s vs. 52 s, p ≤ 0.05), and required more needle redirections (MS1: 4.4 vs. 2.0, p > 0.5; MS4: 4.8 vs. 2.8, p > 0.5; PGY1: 4.4 vs. 2.8, p > 0.5; PGY3: 2.0 vs. 1.0, p > 0.5). CONCLUSIONS: In this study, it was possible to perform ultrasound-guided procedures with Google Glass. Google Glass wearers, on average, took longer to gain access, and had more needle redirections, but less head movements were noted.

Constructing synthetic datasets with generative artificial intelligence to train large language models to classify acute renal failure from clinical notes.

OBJECTIVES: To compare performances of a classifier that leverages language models when trained on synthetic versus authentic clinical notes. MATERIALS AND METHODS: A classifier using language models was developed to identify acute renal failure. Four types of training data were compared: (1) notes from MIMIC-III; and (2, 3, and 4) synthetic notes generated by ChatGPT of varied text lengths of 15 (GPT-15 sentences), 30 (GPT-30 sentences), and 45 (GPT-45 sentences) sentences, respectively. The area under the receiver operating characteristics curve (AUC) was calculated from a test set from MIMIC-III. RESULTS: With RoBERTa, the AUCs were 0.84, 0.80, 0.84, and 0.76 for the MIMIC-III, GPT-15, GPT-30- and GPT-45 sentences training sets, respectively. DISCUSSION: Training language models to detect acute renal failure from clinical notes resulted in similar performances when using synthetic versus authentic training data. CONCLUSION: The use of training data derived from protected health information may not be needed.

Association of amisulpride and recovery room length of stay among patients with postoperative nausea and vomiting following outpatient surgery.

STUDY OBJECTIVE: Postoperative nausea and vomiting (PONV) is a common sequela of surgery in patients undergoing general anesthesia. Amisulpride has shown promise in its ability to treat PONV. The objective of this study was to determine if amisulpride is associated with significant changes in PACU efficiency within a fast-paced ambulatory surgery center. METHODS: This was a retrospective cohort study of 816 patients at a single ambulatory surgery center who experienced PONV between 2018 and 2023. The two cohorts analyzed were patients who did or did not have amisulpride among their anti-emetic regimens in the PACU during two distinct time periods (before and after amisulpride was introduced). The primary outcome of the study was PACU length of stay. Both unmatched analysis and a linear multivariable mixed-effects model fit by restricted maximum likelihood (random effect being surgical procedure) were used to analyze the association between amisulpride and PACU length of stay. We performed segmented regression to account for cohorts occurring during two time periods. RESULTS: Unmatched univariate analysis revealed no significant difference in PACU length of stay (minutes) between the amisulpride and no amisulpride cohorts (115 min vs 119 min, respectively; P = 0.07). However, when addressing confounders by means of the mixed-effects multivariable segmented regression, the amisulpride cohort was associated with a statistically significant reduction in PACU length of stay by 26.1 min (P < 0.001). CONCLUSIONS: This study demonstrated that amisulpride was associated with a significant decrease in PACU length of stay among patients with PONV in a single outpatient surgery center. The downstream cost-savings and operational efficiency gained from this drug's implementation may serve as a useful lens through which this drug's widespread implementation may further be rationalized.

Obesity and outcomes in patients undergoing upper airway surgery for obstructive sleep apnea.

OBJECTIVE: Obesity is frequently debated as a factor associated with increased postoperative complications. Specifically, upper airway surgeries for obstructive sleep apnea (OSA), a common comorbidity among obese patients, may be complicated by obesity's impact on intraoperative ventilation. The aim of this retrospective study was to analyze the association of various degrees of obesity with postoperative outcomes in patients undergoing surgery for OSA. METHODS: The American College of Surgeons National Surgical Quality Improvement database between 2015 and 2019 was used to create a sample of patients diagnosed with OSA who underwent uvulopalatopharyngoplasty, tracheotomy, and surgeries at the base of tongue, maxilla, palate, or nose/turbinate. Inverse probability-weighted logistic regression and unadjusted multivariable logistic regression were used to compare outcomes of non-obese and obesity class 1, class 2, and class 3 groups (World Health Organization classification). Primary outcome was a composite of 30-day readmissions, reoperations, and/or postoperative complications, and a secondary outcome was all-cause same-day hospital admission. RESULTS: There were 1929 airway surgeries identified. The inverse probability-weighted regression comparing class 1, class 2, and class 3 obesity groups to non-obese patients showed no association between obesity and composite outcome and no association between obesity and hospital admission (all p-values > 0.05). CONCLUSION: These results do not provide evidence that obesity is associated with poorer outcomes or hospital admission surrounding upper airway surgery for OSA. While these data points towards the safety of upper airway surgery in obese patients with OSA, larger prospective studies will aid in elucidating the impact of obesity.