Human vs machine: identifying ChatGPT-generated abstracts in Gynecology and Urogynecology.

BACKGROUND: ChatGPT, a publicly available artificial intelligence large language model, has allowed for sophisticated artificial intelligence technology on demand. Indeed, use of ChatGPT has already begun to make its way into medical research. However, the medical community has yet to understand the capabilities and ethical considerations of artificial intelligence within this context, and unknowns exist regarding ChatGPT's writing abilities, accuracy, and implications for authorship. OBJECTIVE: We hypothesize that human reviewers and artificial intelligence detection software differ in their ability to correctly identify original published abstracts and artificial intelligence-written abstracts in the subjects of Gynecology and Urogynecology. We also suspect that concrete differences in writing errors, readability, and perceived writing quality exist between original and artificial intelligence-generated text. STUDY DESIGN: Twenty-five articles published in high-impact medical journals and a collection of Gynecology and Urogynecology journals were selected. ChatGPT was prompted to write 25 corresponding artificial intelligence-generated abstracts, providing the abstract title, journal-dictated abstract requirements, and select original results. The original and artificial intelligence-generated abstracts were reviewed by blinded Gynecology and Urogynecology faculty and fellows to identify the writing as original or artificial intelligence-generated. All abstracts were analyzed by publicly available artificial intelligence detection software GPTZero, Originality, and Copyleaks, and were assessed for writing errors and quality by artificial intelligence writing assistant Grammarly. RESULTS: A total of 157 reviews of 25 original and 25 artificial intelligence-generated abstracts were conducted by 26 faculty and 4 fellows; 57% of original abstracts and 42.3% of artificial intelligence-generated abstracts were correctly identified, yielding an average accuracy of 49.7% across all abstracts. All 3 artificial intelligence detectors rated the original abstracts as less likely to be artificial intelligence-written than the ChatGPT-generated abstracts (GPTZero, 5.8% vs 73.3%; P<.001; Originality, 10.9% vs 98.1%; P<.001; Copyleaks, 18.6% vs 58.2%; P<.001). The performance of the 3 artificial intelligence detection software differed when analyzing all abstracts (P=.03), original abstracts (P<.001), and artificial intelligence-generated abstracts (P<.001). Grammarly text analysis identified more writing issues and correctness errors in original than in artificial intelligence abstracts, including lower Grammarly score reflective of poorer writing quality (82.3 vs 88.1; P=.006), more total writing issues (19.2 vs 12.8; P<.001), critical issues (5.4 vs 1.3; P<.001), confusing words (0.8 vs 0.1; P=.006), misspelled words (1.7 vs 0.6; P=.02), incorrect determiner use (1.2 vs 0.2; P=.002), and comma misuse (0.3 vs 0.0; P=.005). CONCLUSION: Human reviewers are unable to detect the subtle differences between human and ChatGPT-generated scientific writing because of artificial intelligence's ability to generate tremendously realistic text. Artificial intelligence detection software improves the identification of artificial intelligence-generated writing, but still lacks complete accuracy and requires programmatic improvements to achieve optimal detection. Given that reviewers and editors may be unable to reliably detect artificial intelligence-generated texts, clear guidelines for reporting artificial intelligence use by authors and implementing artificial intelligence detection software in the review process will need to be established as artificial intelligence chatbots gain more widespread use.

A survey of preferences and expectations for surgical interventions targeting atonic seizures in Lennox-Gastaut syndrome.

PURPOSE: To assess preferences and outcome expectations for vagus nerve stimulation (VNS) and corpus callosotomy (CC) surgeries in the treatment of atonic seizure in Lennox-Gastaut syndrome (LGS). METHODS: A total of 260 surveys were collected from patients are caregivers of LGS patients via Research Electronic Data Capture (REDCap). RESULTS: Respondents reported an average acceptable atonic seizure reduction rate of 55.9% following VNS and 74.7% following CC. 21.3% (n = 50) were willing to be randomized. Respondents reported low willingness for randomization and a higher seizure reduction expectation with CC. CONCLUSION: Our findings guide surgical approaches for clinicians to consider patient preference in order to design future studies comparing effectiveness between these two procedures.

Robotic Burch colposuspension: anatomical and technical considerations.

INTRODUCTION AND HYPOTHESIS: Up to 13.6% of women will undergo surgical treatment for stress urinary incontinence during their lifetime. Midurethral slings are the mainstay of stress incontinence treatment; however, diversity of surgical options is needed to serve the large number of patients desiring treatment. The Burch colposuspension remains a viable treatment option for appropriately selected patients. Currently, information on procedural standardization and tools for surgical training on robot-assisted colposuspension is limited. METHODS: We describe a stepwise robotic approach aimed at enhancing procedural reproducibility, while decreasing risks of intraoperative injury and postoperative complications. We analyze perioperative outcomes of our technique in a retrospective cohort of patients who underwent robot-assisted colposuspension at our institution. RESULTS: Seven key procedural steps are defined to optimize safe dissection in the retropubic space and to reduce the potential for surgical complications. These include methods of avoiding bladder, urethral, and neurovascular injury, as well as enhancing adequate suture fixation that prevents urethral obstruction and adverse postoperative urinary and pain-related symptoms. Surgical outcomes for 20 patients are reported and reveal low rates of perioperative complications. CONCLUSION: Robot-assisted colposuspension requires thorough knowledge of the retropubic space and the application of standardized techniques may reduce the risk of injury and optimize procedure efficiency and reproducibility.

Linguistic Differences by Gender in Letters of Recommendation for Maternal-Fetal Medicine Fellowship Applicants.

OBJECTIVE: As in many realms of academia and medicine, in obstetrics and gynecology, women experience gender bias in residency evaluations and academic promotions. More specifically, women in Maternal-Fetal Medicine (MFM) are underrepresented within departmental leadership positions. As a means of identifying spaces where bias may exist, multiple investigators have previously reported on gender bias in letters of recommendation (LORs) for residency and subspecialty training programs. We aimed to determine if linguistic differences exist in LORs for self-identified male and female applicants to MFM fellowship at an academic institution. STUDY DESIGN: This was a retrospective single-site cohort study from 2019 to 2021. Data collected included applicant's age, self-reported race/ethnicity and gender, geographic region of residency, step 1 and 2 scores, scholarly and volunteer activities, and number of LORs. The Linguistic Inquiry and Word Count (LIWC) software, a validated text analysis program, was used to characterize LOR linguistic content. Multivariable analysis was used to compare letter characteristics to applicant demographics. RESULTS: A total of 212 applications were reviewed, including 808 LORs. Women comprised 76.9% of applicants, and men 23.1%. Most applicants identified as non-Hispanic White (52.8%). Men were more likely to be international medical graduates (20 vs. 6%, p ≤ 0.01), and women reported more volunteer activities (7.1 ± 5.1 vs. 5.5 ± 4.3, p = 0.04). There were no differences in step scores, number of research projects, or number of LORs. Multivariable analysis controlling for applicant race, step 1 score, and gender of letter writer revealed that letters written for males contained significantly more references to the word category cognitive processes (7.4 ± 0.2 vs. 7.1 ± 0.1, p = 0.046), specifically in reference to the subcategories of certainty and differentiation. CONCLUSION: We identified linguistic differences in LORs written for MFM applicants, suggesting potential bias in the style of writing for male and female physicians applying to this field. KEY POINTS: · Gender bias exists in the evaluation and promotion of women in medicine.. · We sought to determine whether it also exists in letters of recommendation for MFM fellowship.. · Previous studies have examined gender bias in letters of recommendation for other specialties.. · Linguistic bias was detected in letters of recommendation for MFM fellowship from 2019 through 2021.. · We found evidence of linguistic differences based on gender of applicant and letter writers..

Evaluation and Disposition of Fetal Myelomeningocele Repair Candidates: A Large Referral Center Experience.

BACKGROUND: Fetal myelomeningocele (fMMC) repair yields superior outcomes to postnatal repair and is increasingly offered at select fetal centers. OBJECTIVES: To report the fMMC referral process from initial referral to evaluation and surgical intervention in a large fetal referral center. METHODS: We conducted a retrospective cohort study of patients referred to Texas Children's Fetal Center for fMMC between September 2013 and January 2018, reviewing the process from referral to final disposition. The stepwise evaluation included a phone interview followed by multidisciplinary consultation at our fetal center. We modified the Management of Myelomeningocele Study inclusion and exclusion criteria to allow a maternal body mass index of 35-40 on an individual basis. RESULTS: A total of 204 referrals were contacted for a phone interview; 175 (86%) pursued outpatient evaluation, and 80 (46%) of them qualified for repair. Among the eligible patients, 37 (46%) underwent fetoscopic repair, 20 (25%) underwent open repair, and 17 (21%) declined prenatal surgery. Of the 89 noneligible patients (53%) excluded upon outpatient evaluation, 64 (72%) were excluded for fetal and 17 (19%) for maternal reasons. No hindbrain herniation (16%) and maternal BMI and/or hypertension (5%) were the most common reasons for fetal and maternal exclusion, respectively. A total of 31% of our referral population underwent fetal surgery. CONCLUSIONS: A small percentage of fMMC referrals ultimately undergo prenatal surgery. Stepwise evaluation and multidisciplinary teams are key to the success of large referral programs.

Cold ischemia time is an important risk factor for post-liver transplant prolonged length of stay.

Risk analysis of cold ischemia time (CIT) in liver transplantation has largely focused on patient and graft survival. Posttransplant length of stay is a sensitive marker of morbidity and cost. We hypothesize that CIT is a risk factor for posttransplant prolonged length of stay (PLOS) and aim to conduct an hour-by-hour analysis of CIT and PLOS. We retrospectively reviewed all adult, first-time liver transplants between March 2002 and September 2016 in the United Network for Organ Sharing database. The 67,426 recipients were categorized by hourly CIT increments. Multivariate logistic regression of PLOS (defined as >30 days), CIT groups, and an extensive list of confounding variables was performed. Linear regression between length of stay and CIT as continuous variables was also performed. CIT 1-6 hours was protective against PLOS, whereas CIT >7 hours was associated with increased odds for PLOS. The lowest odds for PLOS were observed with 1-2 hours (odds ratio [OR], 0.65; 95% confidence interval [CI], 0.45-0.92) and 2-3 hours (OR, 0.65; 95% CI, 0.55-0.78) of CIT. OR for PLOS steadily increased with increasing CIT, reaching the greatest odds for PLOS with 13-14 hours (OR, 2.05; 95% CI, 1.57-2.67) and 15-16 hours (OR, 2.06; 95% CI, 1.27-3.33) of CIT. Linear regression revealed a positive correlation between length of stay and CIT with a correlation coefficient of +0.35 (P < 0.001). In conclusion, post-liver transplant length of stay is sensitive to CIT, with a substantial increase in the odds of PLOS observed with nearly every additional hour of cold ischemia. We conclude that CIT should be minimized to protect against the morbidity and cost associated with posttransplant PLOS. Liver Transplantation 24 762-768 2018 AASLD.

Comparison of magnetic resonance-guided laser interstitial thermal therapy corpus callosum ablation to open microsurgical corpus callosotomy: A single-center retrospective cohort study.

OBJECTIVE: Corpus callosotomy (CC) is an important treatment for atonic seizures in patients with generalized or multifocal drug-resistant epilepsy (DRE). Traditionally, CC is performed via an open microsurgical approach, but more recently, MR-guided stereotactic laser interstitial thermal therapy (LITT) corpus callosum ablation (CCA) has been developed to leverage the safety and minimally invasive nature of LITT. Given the recent adoption of CCA at select centers, how CCA compares to CC is unknown. We aim to compare the clinical seizure outcomes of CCA and CC after extended follow-up. METHODS: We performed a retrospective cohort study to compare the effectiveness and safety of CC to CCA from 1994 to 2022. The primary outcome was a 50% reduction in target seizure. Secondary outcome measures were postoperative length of stay, adverse events, and other effectiveness metrics. Comparative statistics were executed using Stata. Normality for continuous variables was assessed, and parametric statistics were utilized as needed. Frequency was compared with chi-squared or Fischer's exact tests, when applicable. RESULTS: Data from 47 operations performed on 36 patients were included in this study, of which 13 (36%) patients underwent 17 CCA. Patients who received CCA had similar rates of meaningful reduction (>50%) of atonic seizures as their CC counterparts (55% vs 70% P = 0.15). Patients undergoing CCA had significantly shorter hospitalizations than those receiving CC (2.5 vs 6.0 days P < 0.001). There was no significant difference in rates of postoperative complications between the groups, although the magnitude of the complication rates was lower in the CCA cohort (12% vs 28%). SIGNIFICANCE: This early experience suggests CCA has similar outcomes to traditional CC, albeit with a shorter hospital stay. However, future studies are necessary to investigate the noninferiority between these two approaches. Large multicenter studies are necessary to investigate differences in adverse events and whether these findings generalize across other centers.

Development of a clinical model to predict vagus nerve stimulation response in pediatric patients with drug-resistant epilepsy.

OBJECTIVE: Epilepsy impacts 470,000 children in the United States. For patients with drug-resistant epilepsy (DRE) and unresectable seizure foci, vagus nerve stimulation (VNS) is a treatment option. Predicting response to VNS has been historically challenging. The objective of this study was to create a clinical VNS prediction tool for use in an outpatient setting. METHODS: The authors performed an 11-year retrospective cohort analysis with 1-year follow-up. Patients < 21 years of age with DRE who underwent VNS (n = 365) were included. Logistic regressions were performed to assess clinical factors associated with VNS response (≥ 50% seizure frequency reduction after 1 year); 70% and 30% of the sample were used to train and validate the multivariable model, respectively. A prediction score was subsequently developed. Sensitivity, specificity, and area under the receiver operating characteristic curve (AUC) were calculated. RESULTS: Variables associated with VNS response were < 4-year epilepsy duration before VNS (p = 0.008) and focal motor seizures (p = 0.037). The variables included in the clinical prediction score were epilepsy duration before VNS, age at seizure onset, number of pre-VNS antiseizure medications, if VNS was the patient's first therapeutic epilepsy surgery, and predominant seizure semiology. The final AUCs were 0.7013 for the "fitted" sample and 0.6159 for the "validation" sample. CONCLUSIONS: The authors developed a clinical model to predict VNS response in a large sample of pediatric patients treated with VNS. Despite the large sample size, clinical variables alone were not able to accurately predict VNS response. This score may be useful after further validation, although its predictive ability underscores the need for more robust biomarkers to predict treatment response.

Linguistic Differences by Gender in Letters of Recommendation for Female Pelvic Medicine and Reconstructive Surgery Fellowship Applicants From 2010 to 2020.

IMPORTANCE: Linguistic differences suggestive of gender bias have been detected in letters of recommendation (LOR) for female and male residency and fellowship program applicants within multiple medical specialties. OBJECTIVE: The aim of the study was to determine whether linguistic differences exist in LOR for female and male physicians applying to female pelvic medicine and reconstructive surgery (FPMRS) fellowship. STUDY DESIGN: A retrospective analysis of FPMRS fellowship applications submitted to a university-affiliated academic center from 2010 to 2020 was performed. Linguistic Inquiry and Word Count, a text analysis software, was used to characterize the linguistic content of letters. Multivariable analysis was used to compare letter characteristics with applicant and letter writer demographics. RESULTS: Of 306 applications reviewed, 221 (72.2%) applicants were female and 85 (27.8%) were male. Of the 1,062 letters analyzed, 457 (43.0%) were written by female letter writers, 586 (55.2%) by males, and 19 (1.8%) were a combination. Multivariable analysis controlling for race, Step 1 score, and letter writer gender demonstrated more frequent use of affiliation words for female applicants compared with males (3.1% ± 0.3 vs. 2.9% ± 0.3, P = 0.02). No additional differences were noted in average letter length or all other linguistic categories analyzed. Multiple differences were detected between writing styles of female and male letter writers, including average letter length, use of multiple word categories, and use of communal (relationship-oriented) language. Data were stratified into 2-year periods and no longitudinal trends in linguistic differences were detected. CONCLUSIONS: No linguistic differences, suggestive of gender bias, were found between female and male applicants to FPMRS fellowship.

Linguistic Differences by Gender in Letters of Recommendation for Minimally Invasive Gynecologic Surgery Fellowship Applicants.

OBJECTIVE: Various surgical specialties have reported gender bias in letters of recommendation (LOR). We aimed to determine if linguistic differences exist in LOR for female and male physicians applying to Fellowship in Minimally Invasive Gynecologic Surgery (FMIGS). DESIGN: Retrospective cohort study including application cycles 2019 and 2020. SETTING: Academic university hospital. PARTICIPANTS: FMIGS applicants. RESULTS: Applicant demographic and baseline data included age, race, gender, geographical region of residency training, Step 1 and 2 scores, number of research and volunteer activities, Alpha Omega Alpha and Gold Humanism status and number of LOR, as well as the gender and academic rank of the letter writer. The Linguistic Inquiry and Word Count software, a validated text analysis program, was used to characterize LOR linguistic content. A total of 118 applications, including 391 letters, were analyzed. Seventy-six (64.4%) applicants were female and 42 (35.6%) were male. Most female applicants were white (46% vs. 36%, p = 0.04), had Alpha Omega Alpha status (13% vs. 0%, p = 0.01), higher Step 2 scores (239.7 vs. 230.4, p < 0.01), and more service activities (7.7 vs. 4.7, p < 0.01), compared to male applicants. Male applicants were more likely to graduate from international medical schools (45% vs. 16%, p < 0.01). Female authors wrote 159 LOR, and male authors wrote 232. Following multivariable analysis controlling for race, Step 1 score and letter writer gender, no significant differences in average LOR word count for female and male applicants (406.7 ± 24.2 words vs. 340.1 ± 35.4 words), or differences in Linguistic Inquiry and Word Count linguistic categories existed. CONCLUSIONS: Although Baseline differences were noted between female and male FMIGS applicants, no differences in LOR length or linguistic categories were noted. These results likely reflect the impact of female predominance and increased efforts to address gender bias within Obstetrics and Gynecology.

Liver transplantation as definitive treatment of an unresectable mesenchymal hamartoma in a child with Beckwith-Wiedemann Syndrome.

Mesenchymal liver hamartomas are benign tumors that can cause life-threatening abdominal distension and carry a risk for malignant transformation. In this case report, we describe a 13-month-old male with Beckwith-Wiedemann Syndrome (BWS) who presented with multiple mesenchymal liver hamartomas causing severe intra-abdominal mass effect. Imaging revealed six large multi-locular cystic lesions, ranging from 3.8 to 8.9 cm in diameter. The large size and spread of the tumors necessitated liver transplantation for complete removal. The patient successfully underwent cadaveric piggyback liver transplantation at 25 months of age. He was alive at 16-month follow-up without evidence of tumor recurrence or graft rejection. Histological examination of the hepatic masses revealed mucinous epithelial lining and abundant hepatocytes in varying stages of differentiation, supporting the diagnosis of mesenchymal hamartoma. To the best of our knowledge, this is the first reported case of liver transplantation in a patient with BWS as definitive treatment for unresectable mesenchymal liver hamartoma.