Artificial Intelligence-enabled Decision Support in Surgery: State-of-the-art and Future Directions.

OBJECTIVE: To summarize state-of-the-art artificial intelligence-enabled decision support in surgery and to quantify deficiencies in scientific rigor and reporting. BACKGROUND: To positively affect surgical care, decision-support models must exceed current reporting guideline requirements by performing external and real-time validation, enrolling adequate sample sizes, reporting model precision, assessing performance across vulnerable populations, and achieving clinical implementation; the degree to which published models meet these criteria is unknown. METHODS: Embase, PubMed, and MEDLINE databases were searched from their inception to September 21, 2022 for articles describing artificial intelligence-enabled decision support in surgery that uses preoperative or intraoperative data elements to predict complications within 90 days of surgery. Scientific rigor and reporting criteria were assessed and reported according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews guidelines. RESULTS: Sample size ranged from 163-2,882,526, with 8/36 articles (22.2%) featuring sample sizes of less than 2000; 7 of these 8 articles (87.5%) had below-average (<0.83) area under the receiver operating characteristic or accuracy. Overall, 29 articles (80.6%) performed internal validation only, 5 (13.8%) performed external validation, and 2 (5.6%) performed real-time validation. Twenty-three articles (63.9%) reported precision. No articles reported performance across sociodemographic categories. Thirteen articles (36.1%) presented a framework that could be used for clinical implementation; none assessed clinical implementation efficacy. CONCLUSIONS: Artificial intelligence-enabled decision support in surgery is limited by reliance on internal validation, small sample sizes that risk overfitting and sacrifice predictive performance, and failure to report confidence intervals, precision, equity analyses, and clinical implementation. Researchers should strive to improve scientific quality.

Effect of telehealth-based versus in-person nutritional and exercise intervention on type II diabetes mellitus improvement and efficiency of human resources utilization in patients with obesity.

Aims: Telehealth became a patient necessity during the COVID pandemic and evolved into a patient preference in the post-COVID era. This study compared the % total body weight loss (%TBWL), HbA1c reduction, and resource utilization among patients with obesity and diabetes who participated in lifestyle interventions with or without telehealth. Methods: A total of 150 patients with obesity and diabetes who were followed every 4-6 weeks either in-person (n = 83) or via telehealth (n = 67), were included. All patients were provided with an individualized nutritional plan that included a weight-based daily protein intake from protein supplements and food, an activity/sleep schedule-based meal times, and an aerobic exercise goal of a 2000-calorie burn/week, customized to patient's preferences, physical abilities, and comorbidities. The goal was to lose 10%TBWL. Telehealth-based follow-up required transmission via texting of weekly body composition measurements and any blood glucose levels below 100 mg/dl for medication adjustments. Weight, BMI, %TBWL, HbA1c (%), and medication effect score (MES) were compared. Patient no-show rates, number of visits, program duration, and drop-out rate were used to assess resource utilization based on cumulative staff and provider time spent (CSPTS), provider lost time (PLT) and patient spent time (PST). Results: Mean age was 47.2 ± 10.6 years and 74.6% were women. Mean Body Mass Index (BMI) decreased from 44.1 ± 7.7-39.7 ± 6.7 kg/m2 (p < 0.0001). Mean program duration was 189.4 ± 169.3 days. An HbA1c% unit decline of 1.3 ± 1.5 was achieved with a 10.1 ± 5.1%TBWL. Diabetes was cured in 16% (24/150) of patients. %TBWL was similar in regards to telehealth or in-person appointments (10.6% ± 5.1 vs. 9.6% ± 4.9, p = 0.14). Age, initial BMI, MES, %TBWL, and baseline HbA1c had a significant independent effect on HbA1c reduction (p < 0.0001). Program duration was longer for in-person follow-up (213.8 ± 194 vs. 159.3 ± 127, p = 0.019). The mean annual telehealth and in-person no-show rates were 2.7% and 11.2%, respectively (p < 0.0001). Mean number of visits (5.7 ± 3.0 vs. 8.6 ± 5.1) and drop-out rates (16.49% vs. 25.83%) were lower in telehealth group (p < 0.0001). The CSPTS (440.4 ± 267.5 min vs. 200.6 ± 110.8 min), PLT (28.9 ± 17.5 min vs. 3.1 ± 1.6 min), and PST (1033 ± 628 min vs. 113.7 ± 61.4 min) were significantly longer (p < 0.0001) for the in-person group. Conclusions: Telehealth offered comparable %TBWL and HbA1c decline as in-person follow-up, but with a shorter follow-up, fewer appointments, and no-shows. If improved resource utilization is validated by other studies, telehealth should become the standard of care for the management of obesity and diabetes.

Is Hypertension Despite Use of Two or More Anti-hypertensive Medications for Patients with a BMI < 40 kg/m2 Justified for Eligibility for Bariatric Surgery?

PURPOSE: Several insurance policies require the presence of hypertension, defined as blood pressure > 140 mmHg systolic (SBP), or 90 mmHg diastolic (DBP), despite concurrent use of >1 anti-hypertensive agents for patients with a BMI <40 kg/m2 to qualify for bariatric surgery (BS). The scientific validity of this requirement is assessed. MATERIAL AND METHODS: A total of 461 patients who underwent BS in a community hospital were included. Systolic (SBP) and diastolic (DBP) blood pressure BP (in mmHg) was assessed by an automated manometer at each office visit until 3 years postoperatively and recorded in a prospectively maintained database. RESULTS: Thirty-six (7.8%) patients with BMI <40 Kg/m2, treated by 1, 2, or three anti-hypertensive medications and BP below 140/90, would have been denied BS under such policies. Number of anti-hypertensive medications had no impact on SBP/DBP control preoperatively. Patients being treated preoperatively with < 3 anti-hypertensive medications had a significantly higher percentage of hypertension resolution at 1 (one-med: 66.2%, two-med: 50.9% vs. three-med: 12.5%, p<.0001), 2 (one-med: 63.9%, two-med: 52.8% vs. three-med: 15.4%, p=.0068), and 3 (one-med: 76.9%, two-med: 52.9% vs. three-med: 20%, p=.005) years postoperatively. Multivariate regression demonstrated a significant correlation between the number of preoperative anti-hypertensive medications, preoperative SBP or DBP, and %TBWL. CONCLUSION: Patients treated with fewer than 3 anti-hypertensive medications did not have a better BP control preoperatively but were more likely to resolve their hypertension postoperatively. Restricting access to BS of patients with hypertension treated by fewer than 3 anti-hypertensive medications is not supported by this study.

Prospective Randomized Comparison of Linear Endostaplers During Laparoscopic Sleeve Gastrectomy.

PURPOSE: The development of Laparoscopic Linear Endostaplers (LLES) is crucial in minimally invasive approaches in bariatric surgery, but there have been very few published studies comparing 6-row LLES in Laparoscopic Sleeve Gastrectomy (LSG). The objective of this study was to compare two 6-row LLES in LSG. METHODS: A total of 60 patients were prospectively randomized to undergo LSG with either Medtronic Endo GIA™ Tri-Staple technology (MTS) or AEON ™ Endostapler(Lexington Medical) LLES. The measured parameters included patient demographics, comorbidity indices, LLES and specimen characteristics, postoperative symptoms, hospital stay, and total adverse events (AEs). Intraoperative bleeding was evaluated using five laparoscopic and corresponding endoscopic images of staple line before clip application, compared with a 1-5 Visual Analogue Scale (VAS), assessed by an independent bariatric surgeon who was blinded to the LLES used. Images of all cases were reviewed on the same day to increase test-retest reliability. RESULTS: Both groups were similar in patient demographics. Compared to MTS, AEON LLES group had significantly lower bleeding VAS scores in 4/5 laparoscopic images (pre-pyloric: 1.7 ± 0.7 vs. 2.36 ± 0.76, p = 0.0007, mid-sleeve: 1.46 ± 0.62 vs. 1.86 ± 0.68, p = 0.019, proximal sleeve: 1.6 ± 0.77 vs. 2.0 ± 0.83, p = 0.038, gastro-esophageal junction: 1.43 ± 0.67 vs. 1.86 ± 0.77, p = 0.014) and 3/5 endoscopic images (pre-pyloric: 1.56 ± 0.56 vs. 2.36 ± 0.76, p = 0.006, incisura: 1.66 ± 0.54 vs. 2.0 ± 0.52, p = 0.021, mid-sleeve: 1.63 ± 0.49 vs. 2.0 ± 0.45, p = 0.005). There was no statistical difference in other parameters. CONCLUSION: Both devices were equally safe and effective in terms of LLES and specimen characteristics, patient symptoms, hospital stay, and AEs. Bleeding VAS scores were significantly lower, favoring the AEON LLES.

Advanced age and sex as predictors of adverse outcomes following gastric bypass surgery.

OBJECTIVES: This study aimed to determine whether advanced age or sex was predictive of adverse outcomes after Roux-en-Y gastric bypass. METHODS: The Pennsylvania State Discharge Database was searched for records of morbidly obese patients who underwent Roux-en-Y gastric bypass. The SASs MIXED Procedure was used to test whether mortality alone or adverse outcomes (postoperative complications, nonroutine hospital transfer and mortality) were significantly related to sex or advanced age (>50 years). The presence of comorbidities was used as a blocking variable. RESULTS: Between 1999 and 2001, 4,685 patients underwent Roux-en-Y gastric bypass in Pennsylvania, of which 82% were female and 20% were older than 50 years of age. Comorbidities were present in 71% of patients. Twenty-eight deaths (0.6%) and 813 adverse outcomes (17.4%) occurred. Mortality was greater in males than in females (1.2% vs. 0.47%, P<0.05) without comorbid interaction. Mortality did not increase with age. Adverse outcomes were related to both sexes (24% male, 16% female, P<0.05) and age (< or = 50, 16% vs. > 50, 23%, P<0.05) with a small comorbid interaction. CONCLUSION: Adverse outcomes are more frequent among males and older patients and are influenced by comorbidities. Male patients have a higher mortality that was not affected by the presence of comorbidities.

Improvement of hypothyroidism after laparoscopic Roux-en-Y gastric bypass for morbid obesity.

BACKGROUND: Laparoscopic Roux-en-Y gastric bypass (LRYGBP) has been very effective in managing a broad range of morbid obesity-related co-morbidities. We report a beneficial effect of LRYGBP that has not been previously observed. METHODS: Between December 1999 and September 2002, 224 patients underwent LRYGBP. Preoperative assessment for hypothyroidism and follow-up data were prospectively collected in our database. Improved thyroid function (ITF) or unchanged thyroid function (UTF) was determined by comparison of preoperative and postoperative thyroxine requirements. RESULTS: 23 of 224 patients (10.3%) were treated preoperatively for hypothyroidism. During a median follow-up of 17 months, hypothyroidism was improved in 10/23 patients (43.5%). 2 patients had complete resolution, and the remaining 8 had reduction (14%-50%) of their thyroxine requirements. ITF occurred at a mean follow-up of 8.9 months and at a mean excess weight loss (EWL) of 57%. 6 of the 8 patients (75%) with ITF >or= 25% had EWL >90% at last follow-up, compared to 1 out of 15 patients (6.6%) with UTF or <25% improvement (P =0.001). Comparison of patients with ITF and UTF over time during a 20-month follow-up, showed no significant difference in mean body mass index (BMI) and mean percentage of EWL. CONCLUSION: Improvement of hypothyroidism may be an additional benefit of bariatric surgery that has not been previously reported. Reduction of thyroxine requirements is most likely the result of the decrease in the BMI.

Utility of total mechanical stapled cervical esophagogastric anastomosis after esophagectomy: a comparison to conventional anastomotic techniques.

BACKGROUND: Cervical esophagogastric anastomosis after esophagectomy is often troubled with anastomotic leak resulting in local sepsis, postoperative stricture, and prolonged hospitalization. We compared the anastomotic outcomes and clinical course of esophagectomy patients undergoing total mechanical stapled esophagogastric anastomosis versus a partial handsewn/mechanical stapled cervical anastomotic technique. METHODS: One hundred eighty-one patients underwent transhiatal (N=146) or 3-field (abdomen/chest/neck incisions) (N=35) esophagectomy. A total mechanical stapled anastomosis was accomplished in 125 patients. A handsewn/mechanical stapled anastomosis was performed in 56 patients. The total mechanical stapled anastomosis was accomplished by using the endoscopic gastrointestinal stapler to construct the posterolateral aspect and a linear stapler to close the anterior aspect of the anastomosis. Total mechanical stapled anastomosis patients had the endoscopic gastrointestinal stapler also used to divide the left gastric vessels and the short gastric mesentery for gastric mobilization. Anastomotic outcomes were analyzed by the leak rate (contrast study) and the need of serial dilations in each group. CONCLUSIONS: Total mechanical stapled technique after esophagectomy with cervical esophagogastric anastomosis appears to be effective in reducing hospitalization and anastomotic complications compared to partial or complete handsewn techniques. Liberal use of endoscopic staplers might shorten operative time. Esophageal surgeons should be aware of the advantages and become skilled with these techniques.

Factors affecting quality of life after minimally invasive Heller myotomy for achalasia.

The effect of prior nonoperative treatment, type of fundoplication, and surgical approach on quality of life after minimally invasive Heller myotomy (MIHM) for achalasia in not known. MIHM for achalasia was performed in 105 patients (primary 102; redo 3). Sixty-five patients had prior nonoperative treatment (dilations in 41; botulinum toxin injections in 13; dilations and botulinum toxin injections in 11). Primary laparoscopic MIHM with fundoplication (Dor in 32; Toupet in 56) was performed in 88 patients and thoracoscopic MIHM without fundoplication in 14. Achalasia and quality-of-life-related symptoms were evaluated prospectively with a visual analogue scoring scale. Median follow-up was 25 months. There was a trend toward a higher incidence of intraoperative esophageal perforation and recurrent dysphagia in patients with prior nonoperative treatment. Patients with prior nonoperative treatment had significant improvement in achalasia-related symptoms postoperatively. Patients with prior botulinum toxin injections with or without dilations had no improvement in quality of life after MIHM. The operative success of MIHM may be compromised if prior nonoperative treatment is used. Botulinum toxin injections may blunt the beneficial effect of MIHM on quality of life. The outcome of MIHM is good regardless of the type of fundoplication or surgical approach.

Robotic camera holder as good as expert camera holder: a randomized crossover trial.

INTRODUCTION: This study aimed to compare the impact of robotic camera holder (RCH) and human camera holder (HCH) on product quality and procedure effectiveness of a simulated laparoscopic procedure. METHODS: This was a prospective randomized crossover trial including voluntary surgical residents. Block randomization generated RCH-HCH or HCH-RCH sequence allocation. The task was suturing a duodenal perforation on foam stomach with intracorporeally knot tying in a simulator. The camera was operated by the same robot and same expert. Product quality was measured by accuracy error, tissue damage, sliding knot, and leak. Procedure effectiveness was measured by operating time, nongoal directed actions, and dangerous actions. Kendall's coefficient tau_b was used for interrater reliability between 2 blinded assessors. RESULTS: Forty-four subjects performed their tasks as allocated. Product quality and procedure effectiveness were similar when first attempt of task was compared with the repeat task by same subject ignoring the type of camera holder. There was no evidence of significant unequal carryover effect when comparison was stratified by RCH-HCH or HCH-RCH sequences. There were no differences in product quality and procedure effectiveness when RCH was compared with HCH. Coefficient tau_b was > or = 0.80 for all but dangerous actions (0.72, P=0.08). CONCLUSIONS: RCH and HCH had similar impact on product quality and procedure effectiveness of simulated laparoscopic procedure.

Recurrence rates after abdominal surgery for complete rectal prolapse: a multicenter pooled analysis of 643 individual patient data.

PURPOSE: This study was designed to determine what impact surgical technique, means of access, and method of rectopexy have on recurrence rates following abdominal surgery for full-thickness rectal prolapse. METHODS: Consecutive individual patient data on age, gender, surgical technique (mobilization-only, mobilization-resection-pexy, or mobilization-pexy), means of access (open or laparoscopic), rectopexy method (suture or mesh), follow-up length, and recurrences were collected from 15 centers performing abdominal surgery for full-thickness rectal prolapse between 1979 and 2001. Recurrence was defined as the presence of full-thickness rectal prolapse after abdominal surgery. Chi-squared test and Cox proportional hazards regression analysis were used to assess statistical heterogeneity. Recurrence-free curves were generated and compared using the Kaplan-Meier method and log-rank test, respectively. RESULTS: Abdominal surgery consisted of mobilization-only (n = 46), mobilization-resection-pexy (n = 130), or mobilization-pexy (n = 467). There were 643 patients. After excluding center 8, there was homogeneity on recurrence rates among the centers with recurrences (n = 8) for age (hazards ratio, 0.6; 95 percent confidence interval, 0.2-1.7; P = 0.405), gender (hazards ratio, 0.6; 95 percent confidence interval, 0.1-2.3; P = 0.519), and center (hazards ratio, 0.3; 95 percent confidence interval, 0.1-1.5; P = 0.142). However, there was heterogeneity between centers with (n = 8) and without recurrences (n = 6) for gender (P = 0.0003), surgical technique (P < 0.0001), means of access (P = 0.01), and rectopexy method (P < 0.0001). The median length of follow-up of individual centers varied from 4 to 127 months (P < 0.0001). There were 38 recurrences at a median follow-up of 43 (range, 1-235) months. The pooled one-, five-, and ten-year recurrence rates were 1.06, 6.61, and 28.9 percent, respectively. Age, gender, surgical technique, means of access, and rectopexy method had no impact on recurrence rates. CONCLUSIONS: Although this study is likely underpowered, the impact of mobilization-only on recurrence rates was similar to that of other surgical techniques.