Assistive applications of artificial intelligence in ophthalmology.

PURPOSE OF REVIEW: Assistive (nonautonomous) artificial intelligence (AI) models designed to support (rather than function independently of) clinicians have received increasing attention in medicine. This review aims to highlight several recent developments in these models over the past year and their ophthalmic implications. RECENT FINDINGS: Artificial intelligence models with a diverse range of applications in ophthalmology have been reported in the literature over the past year. Many of these systems have reported high performance in detection, classification, prognostication, and/or monitoring of retinal, glaucomatous, anterior segment, and other ocular pathologies. SUMMARY: Over the past year, developments in AI have been made that have implications affecting ophthalmic surgical training and refractive outcomes after cataract surgery, therapeutic monitoring of disease, disease classification, and prognostication. Many of these recently developed models have obtained encouraging results and have the potential to serve as powerful clinical decision-making tools pending further external validation and evaluation of their generalizability.

Using Frailty Measures to Predict Functional Outcomes and Mortality After Type II Odontoid Fracture in Elderly Patients: A Retrospective Cohort Study.

STUDY DESIGN: Single-center retrospective cohort study. OBJECTIVES: Type II odontoid fractures occur disproportionately among elderly populations and cause significant morbidity and mortality. It is a matter of debate whether these injuries are best managed surgically or conservatively. Our goal was to identify how treatment modalities and patient characteristics correlated with functional outcome and mortality. METHODS: We identified adult patients (>60 years) with traumatic type II odontoid fractures. We used multivariate regression controlling for patient demographics, Glasgow Coma Scale (GCS) score, Charlson Comorbidity Index (CCI), modified Rankin Scale (mRS) score, modified Frailty Index (mFI-5 and mFI-11), fracture displacement, and conservative vs operative treatment. RESULTS: Of the 59 patients (mean age 77.9 years), 24 underwent surgical intervention and 35 underwent conservative management. Operatively managed patients were younger (73.4 vs 80.6 years, P < .001) and had higher degree of fracture displacement (3.5 vs 1.0 mm, P = .002) than conservatively managed patients but no other differences in baseline characteristics. Twenty-four patients (40.7%) died within the study period (median time to death: 376 days). There were no differences between treatment groups in functional outcomes (mRS or Frankel Grade) or mortality (33.3% in operative group vs 45.7%, P = .34). There was a statistically significant correlation between higher presentation mRS score and subsequent mortality on multivariate analysis (OR = 2.06, 95% CI 1.04-4.10, P = .039), whereas surgical intervention, age, GCS score, CCI, mFI-5, mFI-11, sex, and fracture displacement were not significantly correlated. CONCLUSIONS: Mortality after type II odontoid fractures in elderly patients is common. mRS score at presentation may help predict mortality more accurately than other patient factors.

Pediatric Traumatic Brain Injury in a Geographically Dispersed Population: A Relationship Between Distance to Definitive Neurosurgical Treatment and Outcome.

BACKGROUND: There are limited data on the association between transport distance and outcomes in pediatric patients with severe traumatic brain injuries (sTBIs), despite children having to travel further to pediatric trauma centers (PTCs). OBJECTIVE: To assess whether distance from a PTC is associated with outcomes in children who undergo cranial surgery after sTBI. METHODS: Children with sTBI who underwent craniectomy/craniotomy at our PTC between 2010 and 2019 were identified retrospectively. Of these 92 patients, 83 sustained blunt injury and underwent surgery within 24 hours. The distance from injury location to PTC was based on injury zip code and calculated as Euclidean distance. Variables associated with transport, including distance, time, and rural-urban disparity, were analyzed for correlation with poor outcome. RESULTS: Of the 83 patients identified, 81 had injury location information. Forty patients were injured within 30 miles and 41 were injured ≥30 miles from the PTC. Injury severity and pediatric trauma scores were not significantly different between groups. Sixty-eight children (82%) had a satisfactory outcome and 10 children (12%) died. There was a nonsignificant association between distance traveled and poor outcome, even when the cohort was stratified into those with subdural hematomas and those with nonabusive injuries. CONCLUSIONS: Regardless of the distance from the PTC at which their injury occurred, most children in this cohort made a moderate to good recovery. Children injured at greater distances from the PTC did not have worse outcomes; however, studies with larger cohorts are needed to more definitively assess prehospital pediatric transport systems in this population.

Impact of modified Frailty Index-11 (mFI-11) on postoperative complications in patients undergoing transsphenoidal resection of pituitary tumors: Analysis of 2006-2014 ACS-NSQIP database.

INTRODUCTION: Frailty is a measure of physiologic reserve that is frequently cited as a predictor of postoperative complications. However, the effect of frailty on patients undergoing a relatively common procedure such as transsphenoidal resection of pituitary tumors (TSRPT) is unknown. Therefore, we sought to explore this relationship using a large, national database. METHODS: The 2006-2014 American College of Surgeons National Surgical Quality Improvement Program database was retrospectively reviewed to identify all patients who underwent TSRPT. Frailty scores were assigned using the established 11-factor modified Frailty Index (mFI-11). Patients were divided into low-frailty and high-frailty groups, based on mFI comorbidities of ≤ 1 and ≥ 2, respectively. Univariable and multivariable analyses were performed to evaluate the impact of frailty on postoperative outcomes and mortality. RESULTS: A total of 993 patients were included in the analysis. The low-frailty group consisted of 825 patients; the high-frailty group comprised 168 patients. In univariable analysis, there were no significant differences in medical (low-frailty 4.8%, high-frailty 8.3%; p = 0.069) and surgical (low-frailty 1.1%, high-frailty 1.2%; p = 1.000) complications; however, the high-frailty group had a higher rate of mortality (3%) when compared with the low-frailty group (0.6%; p = 0.016, OR 4.07, p = 0.044) and longer hospitalization (4.5 ±â€¯7.4 vs. 5.8 ±â€¯6.8 days; p = 0.023). In multivariable analysis, frailty was a predictor of mortality but not complications or reoperation. CONCLUSIONS: Our study shows that frailty, as measured by the mFI-11, does not predict postoperative complications in patients who undergo TSRPT, but greater frailty is correlated with higher mortality and increased hospital length of stay.

Evaluating the utility and quality of large administrative databases in pediatric spinal neurosurgery research.

PURPOSE: The purpose of this study was to assess the quality of articles utilizing large administrative databases to answer questions related to pediatric spinal neurosurgery by quantifying their adherence to standard reporting guidelines. METHODS: A systematic literature search was conducted with search terms including "pediatric" and "neurosurgery," associated neurosurgical diagnoses, and the names of known databases. Study abstracts were reviewed to identify clinical studies involving pediatric populations, spine-related pathology or procedures, and large administrative databases. Included studies were graded using the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) criteria. RESULTS: A total of 28 papers of the initial 1496 identified met inclusion criteria. These papers involved 10 databases and had a mean study period of 11.46 ± 12.27 years. The subjects of these research papers were undergoing treatment of scoliosis (n = 5), spinal cord injury (n = 5), spinal cord tumors (n = 9), and spine surgery in general (n = 9). The mean STROBE score was 19.41 ± 2.02 (out of 22). CONCLUSION: Large administrative databases are commonly used within pediatric spine-related neurosurgical research to cover a broad spectrum of research questions and study topics. The heterogeneity of research to this point encourages the continued use of large databases to better understand treatment and diagnostic trends, perioperative and long-term outcomes, and rare pathologies within pediatric spinal neurosurgery.