Current and future applications of artificial intelligence in surgery: implications for clinical practice and research.

Surgeons are skilled at making complex decisions over invasive procedures that can save lives and alleviate pain and avoid complications in patients. The knowledge to make these decisions is accumulated over years of schooling and practice. Their experience is in turn shared with others, also via peer-reviewed articles, which get published in larger and larger amounts every year. In this work, we review the literature related to the use of Artificial Intelligence (AI) in surgery. We focus on what is currently available and what is likely to come in the near future in both clinical care and research. We show that AI has the potential to be a key tool to elevate the effectiveness of training and decision-making in surgery and the discovery of relevant and valid scientific knowledge in the surgical domain. We also address concerns about AI technology, including the inability for users to interpret algorithms as well as incorrect predictions. A better understanding of AI will allow surgeons to use new tools wisely for the benefit of their patients.

Navigating the Ethical Landmines of ChatGPT: Implications of Intelligent Chatbots in Plastic Surgery Clinical Practice.

ChatGPT is a cutting-edge language model developed by OpenAI with the potential to impact all facets of plastic surgery from research to clinical practice. New applications for ChatGPT are emerging at a rapid pace in both the scientific literature and popular media. It is important for clinicians to understand the capabilities and limitations of these tools before patient-facing implementation. In this article, the authors explore some of the technical details behind ChatGPT: what it is, and what it is not. As with any emerging technology, attention should be given to the ethical and health equity implications of this technology on our plastic surgery patients. The authors explore these concerns within the framework of the foundational principles of biomedical ethics: patient autonomy, nonmaleficence, beneficence, and justice. ChatGPT and similar intelligent conversation agents have incredible promise in the field of plastic surgery but should be used cautiously and sparingly in their current form. To protect patients, it is imperative that societal guidelines for the safe use of this rapidly developing technology are developed.

Air or Saline? A Propensity Score-Matched Analysis on the Effect of Tissue Expander Fill on Complications in Immediate Breast Reconstruction.

BACKGROUND: Tissue expander fill medium and volume have implications for the pressure exerted on mastectomy skin flaps. This study evaluated the influence of initial fill medium (air vs. saline) on complications in immediate breast reconstruction within a propensity score-matched cohort. PATIENTS AND METHODS: Patients undergoing immediate tissue expander-based breast reconstruction with initial intraoperative fill with air were propensity score matched 1:2 to those with saline initial fill based on patient and tissue expander characteristics. Incidence of overall and ischemic complications were compared by fill medium (air vs. saline). RESULTS: A total of 584 patients were included, including 130 (22.2%) with initial fill with air, 377 (64.6%) with initial fill with saline, and 77 (13.2%) with 0 cc of initial fill. After multivariate adjustment, higher intraoperative fill volume was associated with increased risk of mastectomy skin flap necrosis [regression coefficient (RC) 15.7; p = 0.049]. Propensity score matching was then conducted among 360 patients (Air: 120 patients vs. Saline: 240 patients). After propensity score matching, there were no significant differences in the incidences of mastectomy skin flap necrosis, extrusion, reoperation, or readmission between the air and saline cohorts (all p > 0.05). However, initial fill with air was associated with lower incidence of infection requiring oral antibiotics (p = 0.003), seroma (p = 0.004), and nipple necrosis (p = 0.03). CONCLUSIONS: Within a propensity score-matched cohort, initial fill with air was associated with a lower incidence of complications, including ischemic complications after nipple-sparing mastectomy. Initial fill with air and lower fill volumes may be strategies to reducing risk of ischemic complications among high-risk patients.

Two Thumbs Down: A 10-Year Analysis of Crossbow-Related Hand and Digit Injuries in the United States.

BACKGROUND: Modern crossbow is a popular weapon used throughout the United States for outdoor recreation. Crossbow mechanics place the hand and digits at significant risk of injury while shooting or handling these weapons; however, injury patterns have been poorly documented. In this study, the authors use a national database to assess patterns of crossbow-related injuries of the hand and digits. METHODS: A retrospective review of the National Electronic Injury Surveillance System database was conducted to identify crossbow-related hand and digit injuries over the past 10 years. Demographic data, injury timing, anatomical injury location, specific injury diagnosis, and disposition data were collected. RESULTS: A total of 15 460 crossbow-related hand injuries were identified between 2011 and 2021. A strong temporal relationship was identified, with 89% of injuries occurring between the months of August and December. Male patients sustained most injuries (>85%). Body areas injured were the digits (93.2%) and the hand (5.7%). The most common injuries were laceration (n = 7520, 48.6%), fracture (n = 4442, 28.7%), amputation (n = 1341, 8.7%), and contusion/abrasion (n = 957, 6.2%). More than 50% of cases involved injury to the thumb, with approximately 750 thumb amputations recorded during the time period examined. CONCLUSIONS: This study is the first to document the patterns of hand and digit injuries associated with crossbow use on a national scale. These findings have important implications for public health awareness campaigns targeting hunters and support the designation of crossbow "safety wings" as a mandatory feature of crossbow design.

Consequences and Predictors of Prolonged Tissue Expander Duration in Breast Reconstruction.

BACKGROUND: Tissue expanders (TEs) are temporary devices used in breast reconstruction, which are generally removed within 1 year. There is a paucity of data regarding the potential consequences when TEs have longer indwelling times. Thus, we aim to determine whether prolonged TE implantation length is associated with TE-related complications. METHODS: This is a single-center retrospective review of patients who underwent TE placement for breast reconstruction from 2015 to 2021. Complications were compared between patients who had a TE for >1 year and <1 year. Univariate and multivariate regressions were used to evaluate predictors of TE complications. RESULTS: A total of 582 patients underwent TE placement and 12.2% had the expander for >1 year. Adjuvant chemoradiation, body mass index (BMI), overall stage, and diabetes predicted the duration of TE placement (p ≤ 0.006). Rate of return to the operating room was higher in patients who had TEs in place >1 year (22.5 vs 6.1%, p < 0.001). On multivariate regression, prolonged TE duration predicted an infection requiring antibiotics, readmission, and reoperation (p < 0.001). Reasons for longer indwelling times included need for additional chemoradiation (79.4%), TE infections (12.7%), and requesting a break from surgery (6.3%). CONCLUSION: Indwelling TEs for >1 year are associated with higher rates of infection, readmission, and reoperation even when controlling for adjuvant chemoradiation. Patients with diabetes, a higher BMI, advanced cancer stage, and those requiring adjuvant chemoradiation should be advised they may require a TE for a longer time interval prior to final reconstruction.

Variations in Incidence of Trigger Finger and Response to Corticosteroid Injection after Aromatase Inhibitor Therapy for Breast Cancer.

BACKGROUND: Aromatase inhibitors (AIs), such as letrozole and anastrozole, have been demonstrated to have significant musculoskeletal symptoms in patients. The purpose of this study was to evaluate the effect of specific AI medications on the incidence of trigger finger and independent factors affecting treatment outcomes within this population. METHODS: A retrospective chart review was performed at the authors' institution between the years 2014 and 2018 in patients with the diagnosis of breast cancer. This cohort was then sorted based on receiving medication regimens, trigger finger diagnosis, steroid injections, and need for surgical release of trigger finger. RESULTS: A total of 15,144 patients were included for initial review. The overall rate of trigger finger diagnosis was 2.75% in the entire breast cancer population and 4.5% for patients receiving AI therapy. Patients taking letrozole and anastrozole had an increased odds ratio of 2.0 and 1.7, respectively, for developing trigger finger. Patients who switched between letrozole and anastrozole during treatment had a higher rate of failed steroid injection treatment (45.2% versus 23.5%; P = 0.021). Among patients receiving AI treatment diagnosed with trigger finger, diabetes and hemoglobin A1c level greater than 6.5 were associated with significantly increased rates of failed steroid therapy. CONCLUSIONS: Patients receiving AI therapy have an increased incidence of trigger finger. The outcomes of treatment are equivalent between AI and non-AI trigger finger populations. However, steroid therapy is more likely to fail in patients who require switching of regimens because of significant musculoskeletal symptoms. Poorly controlled diabetes was also an independent factor for compromised steroid treatment of trigger finger. CLINICAL QUESTION/LEVEL OF EVIDENCE: Therapeutic, III.

Predicting Final Implant Volume in Two-stage Prepectoral Breast Reconstruction.

Two-stage implant-based breast reconstruction remains the most commonly performed reconstructive modality following mastectomy. Although prior studies have explored the relationship between tissue expander (TE) features and permanent implant (PI) size in subpectoral reconstruction, no such study exists in prepectoral reconstruction. This study aims to identify pertinent TE characteristics and evaluate their correlations with PI size for prepectoral implant-based reconstruction. Methods: This study analyzed patients who underwent two-stage prepectoral tissue expansion for breast reconstruction followed by implant placement. Patient demographics and oncologic characteristics were recorded. TE and PI features were evaluated. Significant predictors for PI volume were identified using linear and multivariate regression analyses. Results: We identified 177 patients and 296 breast reconstructions that met inclusion criteria. All reconstructions were performed in the prepectoral plane with the majority using acellular dermal matrix (93.8%) and primarily silicone implants (94.3%). Mean TE size was 485.4 cm3 with mean initial fill of 245.8 cm3 and mean final fill of 454.4 cm3. Mean PI size was 502.9 cm3 with a differential fill volume (PI-TE) of 11.7 cm3. Multivariate analysis identified significant features for PI size prediction, including TE size (R2 = 0.60; P < 0.0001) and TE final fill volume (R2 = 0.57; P < 0.0001). The prediction expression for TE final fill and TE size was calculated as 26.6 + 0.38*(TE final fill) + 0.61*(TE size). Conclusions: TE size and final expansion volume were significant variables for implant size prediction. With prepectoral implant placement gaining popularity, the predictive formula may help optimize preoperative planning and decision-making in prepectoral reconstructions.

Deep Learning Applications in Surgery: Current Uses and Future Directions.

Deep learning (DL) is a subset of machine learning that is rapidly gaining traction in surgical fields. Its tremendous capacity for powerful data-driven problem-solving has generated computational breakthroughs in many realms, with the fields of medicine and surgery becoming increasingly prominent avenues. Through its multi-layer architecture of interconnected neural networks, DL enables feature extraction and pattern recognition of highly complex and large-volume data. Across various surgical specialties, DL is being applied to optimize both preoperative planning and intraoperative performance in new and innovative ways. Surgeons are now able to integrate deep learning tools into their practice to improve patient safety and outcomes. Through this review, we explore the applications of deep learning in surgery and related subspecialties with an aim to shed light on the practical utilization of this technology in the present and near future.

New Frontiers of Natural Language Processing in Surgery.

The vast and ever-growing volume of electronic health records (EHR) have generated a wealth of information-rich data. Traditional, non-machine learning data extraction techniques are error-prone and laborious, hindering the analytical potential of these massive data sources. Equipped with natural language processing (NLP) tools, surgeons are better able to automate, and customize their review to investigate and implement surgical solutions. We identify current perioperative applications of NLP algorithms as well as research limitations and future avenues to outline the impact and potential of this technology for progressing surgical innovation.

Ethical, Legal, and Financial Considerations of Artificial Intelligence in Surgery.

Machine learning systems have become integrated into some of the most vital decision-making aspects of humanity, including hiring decisions, loan applications, and automobile safety, to name just a few. As applications increase in both gravity and complexity, the data quality and algorithmic interpretability of the systems must rise to meet those challenges. This is especially vital for navigating the nuances of health care, particularly among the high stakes of surgical operations. In addition to inherent ethical challenges of enabling a "black box" system to influence decision-making in patient care, the creation of biased datasets leads to biased algorithms with the power to perpetuate discrimination and reinforce disparities. Transparency and responsibility are paramount to the implementation of artificial intelligence in surgical decision-making and autonomous robotic surgery. Machine learning has been permeating health care across diverse clinical and surgical contexts but continues to face sizable obstacles, including apprehension from patients and providers alike. To integrate the technology fully while upholding standard of care and patient-provider trust, one must acknowledge and address the ethical, financial, and legal implications of using artificial intelligence for patient care.

Comparative Effectiveness of Behavioral Interventions on Quality of Life for Older Adults With Mild Cognitive Impairment: A Randomized Clinical Trial.

Importance: Recommendations to engage in behavioral strategies to combat clinically significant cognitive and behavioral symptoms are routinely given to persons with mild cognitive impairment (MCI). The comparative effectiveness of these behavioral interventions is not well understood. Objective: To compare the incremental effects of combinations of 5 behavioral interventions on outcomes of highest importance to patients with MCI. Design, Setting, and Participants: In this multisite, cluster randomized, multicomponent comparative effectiveness trial, 272 patients from 4 academic medical outpatient centers (Mayo Clinic, Rochester, Minnesota; Mayo Clinic, Scottsdale, Arizona; Mayo Clinic, Jacksonville, Florida; and University of Washington, Seattle) were recruited from September 1, 2014, to August 31, 2016, with last follow-up March 31, 2019. All participants met the National Institute on Aging-Alzheimer's Association criteria for MCI. Interventions: The intervention program was modeled on the Mayo Clinic Healthy Action to Benefit Independence and Thinking (HABIT) program, a 50-hour group intervention conducted during 2 weeks, including memory compensation training, computerized cognitive training, yoga, patient and partner support groups, and wellness education. In our study, 1 of 5 interventions was randomly selected to be withheld for each intervention group. Participants and their partners had 1-day booster sessions at 6 and 12 months after intervention. Main Outcomes and Measures: Quality-of-life measurement of participants with MCI at 12 months was the primary outcome, selected based on the preference rankings of previous program participants. Mood, self-efficacy, and memory-based activities of daily living were also highly ranked. Results: A total of 272 participants (mean [SD] age, 75 [8] years; 160 [58.8%] male and 112 [41.2%] female) were enrolled in this study, with 56 randomized to the no yoga group, 54 to no computerized cognitive training, 52 to no wellness, 53 to no support, and 57 to no memory support system. The greatest effect size for quality of life was between the no computerized cognitive training and no wellness education groups at 0.34 (95% CI, 0.05-0.64). In secondary analyses, wellness education had a greater effect on mood than computerized cognitive training (effect size, 0.53; 95% CI, 0.21-0.86), and yoga had a greater effect on memory-related activities of daily living than support groups (effect size, 0.43; 95% CI, 0.13-0.72). Conclusions and Relevance: These results provide further support for behavioral interventions for persons with MCI. Different outcomes were optimized by different combinations of interventions. These findings provide an initial exploration of the effect of behavioral interventions on patient-advocated outcomes in persons with MCI. Trial Registration: ClinicalTrials.gov identifier: NCT02265757.

Physical exercise and cognitive engagement outcomes for mild neurocognitive disorder: a group-randomized pilot trial.

BACKGROUND: Amnestic mild cognitive impairment (aMCI) is considered a risk state for the development of dementia due to Alzheimer's disease. It is also a period in which interventions may be most effective in slowing progression to dementia. Computerized cognitive training and increased physical activity have shown to be among the most promising interventions. However, current evidence from randomized controlled trials comparing cognitive training, physical activity, and an active control is inconsistent. Furthermore, the neural mechanisms underlying these interventions are currently unclear. METHODS: The objective of the current pilot study is to explore the feasibility of a trial investigating the impact of computerized cognitive training, yoga, and an active control intervention (wellness education) in individuals with aMCI by conducting a group-randomized, multisite, parallel, three-arm pilot study. We will establish preliminary effect sizes regarding the association of each intervention with neuroimaging and cognitive and participant-reported measures. We also aim to estimate the strength of association between the various outcomes. The current trial aims to recruit 75 people with aMCI and their 75 cognitively healthy care partners through clinics and senior care facilities. The initial intervention will last 10 days and will consist of 1 h daily of the assigned intervention i.e., (yoga, computerized cognitive training, or wellness education) combined with 1 h of memory compensation training and 1 h of support groups. Twenty-five participants will be group-randomized to each arm using a random number generator. Study staff and participants will be kept blind until recruitment is complete for each group. After the initial two-week intervention, participants will continue the assigned intervention for 24 weeks. Outcome measures are: functional connectivity and cerebral perfusion as assessed by magnetic resonance imaging; cognition; daily functioning; mood; anxiety; self-efficacy; caregiver burden; quality of life; and study feasibility including recruitment and retention rates. DISCUSSION: This pilot trial aims to investigate the feasibility of a trial studying the impact of computerized cognitive training, yoga, and an active control intervention in persons with aMCI on MRI-based functional connectivity and cerebral perfusion as well as cognition, daily functioning, mood, anxiety, and quality of life and feasibility? TRIAL REGISTRATIONS: ClinicalTrials.gov, NCT03095170 . Registered on 23 March 2017.

Behavioral Interventions to Prevent or Delay Dementia: Protocol for a Randomized Comparative Effectiveness Study.

BACKGROUND: Currently, people at risk for dementia and their caregivers are confronted with confusing choices about what behavioral interventions are most effective. OBJECTIVE: The objective of this study is to determine which empirically supported behavioral interventions most impact the outcomes highly valued by patients with mild cognitive impairment and their partners. METHODS: This protocol describes a comparative effectiveness trial targeting 300 participants with mild cognitive impairment and their study partners. The trial is being conducted at the Mayo Clinic campuses in Arizona, Florida, Minnesota, and the University of Washington in Seattle. The study examines the contribution of five behavioral interventions (yoga, memory compensation training, computerized cognitive training, support groups, and wellness education) on primary outcomes of participant and partner quality of life and self-efficacy. In this unique 10-day multicomponent intervention, groups of couples were randomized to have one of the five interventions withheld while receiving the other four. Although the longitudinal follow-up is still under way, enrollment results are available and reported. RESULTS: In total, 272 couples have been enrolled in the trial and follow-up visits continue. Outcomes will be assessed at the end-of-intervention and 6-, 12-, and 18-month follow-ups. We anticipate reporting on our primary and secondary outcomes across time points in the next 2 years. CONCLUSIONS: This paper describes the protocol for a randomized comparative effectiveness study of behavioral interventions to prevent or delay dementia. We describe of the rationale, design, power analysis, and analysis plan. Also because enrollment is complete and we are in follow-up phases of the study, we have included enrollment data from the trial. TRIAL REGISTRATION:  ClinicalTrials.gov NCT02265757; http://clinicaltrials.gov/ctsshow/ NCT02265757 (Archived by WebCite at http://www.webcitation.org/6ueRfwSYv).

Out of Arabia-the settlement of island Soqotra as revealed by mitochondrial and Y chromosome genetic diversity.

The Soqotra archipelago is one of the most isolated landmasses in the world, situated at the mouth of the Gulf of Aden between the Horn of Africa and southern Arabia. The main island of Soqotra lies not far from the proposed southern migration route of anatomically modern humans out of Africa approximately 60,000 years ago (kya), suggesting the island may harbor traces of that first dispersal. Nothing is known about the timing and origin of the first Soqotri settlers. The oldest historical visitors to the island in the 15th century reported only the presence of an ancient population. We collected samples throughout the island and analyzed mitochondrial DNA and Y-chromosomal variation. We found little African influence among the indigenous people of the island. Although the island population likely experienced founder effects, links to the Arabian Peninsula or southwestern Asia can still be found. In comparison with datasets from neighboring regions, the Soqotri population shows evidence of long-term isolation and autochthonous evolution of several mitochondrial haplogroups. Specifically, we identified two high-frequency founder lineages that have not been detected in any other populations and classified them as a new R0a1a1 subclade. Recent expansion of the novel lineages is consistent with a Holocene settlement of the island approximately 6 kya.

Peritraumatic responses and their relationship to perceptions of threat in female crime victims.

Peritraumatic responses, aside from dissociation, have been understudied in acute trauma populations. Participants were 172 female rape, 68 assault, and 80 robbery victims recruited through formal reporting agencies and assessed 1 month after the crime. Despite substantial overlap across crimes, rape victims reported more emotional responses reflecting fear, detachment, shame, and more nonactive behavioral responses. Regression analysis examining the prediction of perceived threat by peritraumatic responses and crime variables indicated that increased duration of crimes; decreased calmness; increased fear; numbing; use of begging, pleading, and crying; and attempts to reason with the perpetrator(s) were all significantly associated with increased appraisal of threat.