-
Frontiers in Medicine 2022Artificial Intelligence (AI) holds considerable promise for diagnostics in the field of gastroenterology. This systematic review and meta-analysis aims to assess the...
BACKGROUND
Artificial Intelligence (AI) holds considerable promise for diagnostics in the field of gastroenterology. This systematic review and meta-analysis aims to assess the diagnostic accuracy of AI models compared with the gold standard of experts and histopathology for the diagnosis of various gastrointestinal (GI) luminal pathologies including polyps, neoplasms, and inflammatory bowel disease.
METHODS
We searched PubMed, CINAHL, Wiley Cochrane Library, and Web of Science electronic databases to identify studies assessing the diagnostic performance of AI models for GI luminal pathologies. We extracted binary diagnostic accuracy data and constructed contingency tables to derive the outcomes of interest: sensitivity and specificity. We performed a meta-analysis and hierarchical summary receiver operating characteristic curves (HSROC). The risk of bias was assessed using Quality Assessment for Diagnostic Accuracy Studies-2 (QUADAS-2) tool. Subgroup analyses were conducted based on the type of GI luminal disease, AI model, reference standard, and type of data used for analysis. This study is registered with PROSPERO (CRD42021288360).
FINDINGS
We included 73 studies, of which 31 were externally validated and provided sufficient information for inclusion in the meta-analysis. The overall sensitivity of AI for detecting GI luminal pathologies was 91.9% (95% CI: 89.0-94.1) and specificity was 91.7% (95% CI: 87.4-94.7). Deep learning models (sensitivity: 89.8%, specificity: 91.9%) and ensemble methods (sensitivity: 95.4%, specificity: 90.9%) were the most commonly used models in the included studies. Majority of studies ( = 56, 76.7%) had a high risk of selection bias while 74% ( = 54) studies were low risk on reference standard and 67% ( = 49) were low risk for flow and timing bias.
INTERPRETATION
The review suggests high sensitivity and specificity of AI models for the detection of GI luminal pathologies. There is a need for large, multi-center trials in both high income countries and low- and middle- income countries to assess the performance of these AI models in real clinical settings and its impact on diagnosis and prognosis.
SYSTEMATIC REVIEW REGISTRATION
[https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=288360], identifier [CRD42021288360].
PubMed: 36405592
DOI: 10.3389/fmed.2022.1018937 -
JAMA Network Open Mar 2021Obstructive sleep apnea (OSA) is a highly prevalent global health concern and is associated with many adverse outcomes for patients. (Meta-Analysis)
Meta-Analysis
IMPORTANCE
Obstructive sleep apnea (OSA) is a highly prevalent global health concern and is associated with many adverse outcomes for patients.
OBJECTIVE
To evaluate the utility of the STOP-Bang (snoring, tiredness, observed apnea, blood pressure, body mass index, age, neck size, gender) questionnaire in the sleep clinic setting to screen for and stratify the risk of OSA among populations from different geographical regions.
DATA SOURCES AND STUDY SELECTION
MEDLINE, MEDLINE In-process, Embase, EmCare Nursing, Cochrane Central Register of Controlled Trials, Cochrane Database of Systematic Reviews, PsycINFO, Journals@Ovid, Web of Science, Scopus, and CINAHL electronic databases were systematically searched from January 2008 to March 2020. This was done to identify studies that used the STOP-Bang questionnaire and polysomnography testing in adults referred to sleep clinics.
DATA EXTRACTION AND SYNTHESIS
Clinical and demographic data were extracted from each article independently by 2 reviewers. The combined test characteristics were calculated using 2 × 2 contingency tables. Random-effects meta-analyses and metaregression with sensitivity analyses were performed. The Preferred Reporting Items for Systematic Review and Meta-analyses (PRISMA) guideline was followed.
MAIN OUTCOMES AND MEASURES
The combined test characteristics and area under summary receiver operating characteristic curves (AUCs) were used to compare STOP-Bang questionnaire accuracy with polysomnography testing.
RESULTS
A total of 47 studies with 26 547 participants (mean [SD] age, 50 [5] years; mean [SD] body mass index, 32 [3]; 16 780 [65%] men) met the criteria for the systematic review. Studies were organized in different geographic regional groups: North America, South America, Europe, Middle East, East Asia, and South or Southeast Asia. The prevalence rates for all OSA, moderate to severe OSA, and severe OSA were 80% (95% CI, 80%-81%), 58% (95% CI, 58%-59%), and 39% (95% CI, 38%-39%), respectively. A STOP-Bang score of at least 3 had excellent sensitivity (>90%) and high discriminative power to exclude moderate to severe and severe OSA, with negative predictive values of 77% (95% CI, 75%-78%) and 91% (95% CI, 90%-92%), respectively. The diagnostic accuracy of a STOP-Bang score of at least 3 to detect moderate to severe OSA was high (>0.80) in all regions except East Asia (0.52; 95% CI, 0.48-0.56).
CONCLUSIONS AND RELEVANCE
The results of this meta-analysis suggest that the STOP-Bang questionnaire can be used as a screening tool to assist in triaging patients with suspected OSA referred to sleep clinics in different global regions.
Topics: Americas; Asia; Europe; Humans; Sleep Apnea, Obstructive; Surveys and Questionnaires; Symptom Assessment
PubMed: 33683333
DOI: 10.1001/jamanetworkopen.2021.1009 -
Critical Care Medicine Jan 2023Evaluate associations between ultrasound measures and difficult laryngoscopy. (Meta-Analysis)
Meta-Analysis
OBJECTIVES
Evaluate associations between ultrasound measures and difficult laryngoscopy.
DATA SOURCES
MEDLINE, Embase, Google Scholar, Web of Science, and the Cochrane Library were searched using MeSH terms and keywords.
STUDY SELECTION
Studies published in English describing the use of airway ultrasound for identifying difficult laryngoscopy, with sufficient data to calculate sensitivity and specificity using 2 × 2 tables.
DATA EXTRACTION
We assigned the described indices of airway dimension to one of three domains based on methodology characteristics: anterior tissue thickness domain, anatomical position domain, and oral space domain. We then performed a bivariate random-effects meta-analysis, deriving pooled sensitivity, specificity, diagnostic odds ratio, positive likelihood ratio, and negative likelihood ratio estimates. We assessed risks of bias using Quality Assessment of Diagnostic Accuracy Studies-2 analysis.
DATA SYNTHESIS
Thirty-three studies evaluating 27 unique indices were included in the meta-analysis. The ultrasound protocols of the included studies were heterogeneous. Anterior tissue thickness demonstrated a pooled sensitivity of 76% (95% CI, 71-81%), specificity of 77% (95% CI, 72-81%), and an area under the receiver operating characteristic curve (AUROC) of 0.83 (95% CI, 0.80-0.86). Anatomical position demonstrated a pooled sensitivity of 74% (95% CI, 61-84%), specificity of 86% (95% CI, 78-91%), and an AUROC of 0.87 (95% CI, 0.84-0.90). Oral space demonstrated a pooled sensitivity of 53% (95% CI, 0.36-0.69), specificity of 77% (95% CI, 0.67-0.85), and an AUROC of 0.73 (95% CI, 0.69-0.77).
CONCLUSIONS
Airway ultrasound metrics associate with difficult laryngoscopy in three domains: anterior tissue thickness, anatomic position, and oral space. An assessment instrument combining clinical and ultrasound assessments may be an accurate screening tool for difficult laryngoscopy.
Topics: Laryngoscopy; Sensitivity and Specificity; Ultrasonography; ROC Curve
PubMed: 36519985
DOI: 10.1097/CCM.0000000000005711 -
Surgical Infections Apr 2018Peri-prosthetic joint infection (PJI) is a serious and frequent complication of total joint arthroplasty (TJA). Recently, synovial fluid leukocyte esterase (LE),... (Meta-Analysis)
Meta-Analysis Review
BACKGROUND
Peri-prosthetic joint infection (PJI) is a serious and frequent complication of total joint arthroplasty (TJA). Recently, synovial fluid leukocyte esterase (LE), measurement of which is convenient and fast, has been examined as a marker of PJI. We summarized the articles describing synovial fluid LE as a biomarker for the diagnosis of PJI and assessed its diagnostic value in patients suspected of having PJI.
METHODS
We searched with appropriate key words in PubMed, Embase, Web of Science, the Cochrane database, and Science Direct. Eligible studies providing sufficient data to construct 2 × 2 contingency tables were chosen on the basis of several criteria, and the quality of the chosen studies was assessed. The pooled sensitivity, specificity, and diagnostic odds ratio (DOR) were calculated for those studies. The summary receiver operating characteristic (SROC) curve and the area under the SROC (AUSROC) were used to evaluate the overall diagnostic performance of LE.
RESULTS
Eleven studies were found suitable for this systematic review. Among them, eight articles with a total of 1,011 participants qualified for meta-analysis. The pooled sensitivity, specificity, and DOR were 0.90 (95% confidence interval [CI] 0.76-0.96), 0.97 (95% CI 0.95-0.98), and 310.76 (95% CI 103.86-929.88), respectively. The SROC was 0.98 (95% CI 0.96-0.99). Sub-group analysis indicated that the sample inclusion criteria might be the main source of heterogeneity. Publication bias was suggested by an asymmetrical funnel plot (p = 0.144).
CONCLUSION
Although the result of synovial fluid LE assay can be influenced by sample-related factors, it is more specific as a means to exclude PJI.
Topics: Adult; Aged; Aged, 80 and over; Carboxylic Ester Hydrolases; Female; Humans; Male; Middle Aged; Prosthesis-Related Infections; Synovial Fluid; Young Adult
PubMed: 29099342
DOI: 10.1089/sur.2017.192 -
The Cochrane Database of Systematic... Aug 2023Liver resection is the optimal treatment for selected benign and malignant liver tumours, but it can be associated with significant blood loss. Numerous anaesthetic and... (Review)
Review
BACKGROUND
Liver resection is the optimal treatment for selected benign and malignant liver tumours, but it can be associated with significant blood loss. Numerous anaesthetic and surgical techniques have been developed to reduce blood loss and improve perioperative outcomes. One such technique is the application of topical fibrin-based haemostatic agents (FBHAs) to the resection surface. There is no standard practice for FBHA use, and a variety of commercial agents and devices are available, as well as non-FBHAs (e.g. collagen-based agents). The literature is inconclusive on the effectiveness of these methods and on the clinical benefits of their routine use.
OBJECTIVES
To evaluate the benefits and harms of fibrin-based haemostatic agents in reducing intraoperative blood loss in adults undergoing liver resection.
SEARCH METHODS
We searched the Cochrane Hepato-Biliary Group (CHBG) Controlled Trials Register, CENTRAL, MEDLINE, Embase, LILACS, Science Citation Index Expanded, and Conference Proceedings Citation Index-Science up to 20 January 2023. We also searched online trial registries, checked the reference lists of all primary studies, and contacted the authors of included trials for additional published or unpublished trials.
SELECTION CRITERIA
We considered for inclusion all randomised clinical trials evaluating FBHAs versus no topical intervention or non-FBHAs, irrespective of publication type, publication status, language of publication, and outcomes reported. Eligible participants could have any liver pathology and be undergoing major or minor liver resections through open or laparoscopic surgery.
DATA COLLECTION AND ANALYSIS
Two review authors independently screened the results of the literature search and used data extraction forms to collate the results. We expressed dichotomous outcome results as risk ratios (RRs) and continuous outcome results as mean differences (MDs), each with their corresponding 95% confidence interval (CI). We used a random-effects model for the main analyses. Our primary outcomes were perioperative mortality, serious adverse events, haemostatic efficacy, and health-related quality of life. Our secondary outcomes were efficacy as sealant, adverse events considered non-serious, operating time, and length of hospital stay. We assessed the certainty of the evidence with GRADE and presented results in two summary of findings tables.
MAIN RESULTS
We included 22 trials (2945 participants) evaluating FBHAs versus no intervention or non-FBHAs; 19 trials with 2642 participants provided data for the meta-analyses. Twelve trials reported commercial funding, one trial reported no financial support, and nine trials provided no information on funding. Below we present the most clinically relevant outcome results, also displayed in our summary of findings table. Fibrin-based haemostatic agents versus no intervention Six trials (1001 participants) compared FBHAs with no intervention. One trial was at low risk of bias in all five domains, and all other trials were at high or unclear risk of bias in at least one domain. Two trials were at high risk of bias related to blinding. It is unclear if FBHAs compared with no intervention have an effect on perioperative mortality (RR 2.58, 95% CI 0.89 to 7.44; 4 trials, 782 participants), serious adverse events (RR 0.96, 95% CI 0.88 to 1.05; 4 trials, 782 participants), postoperative transfusion (RR 1.04, 95% CI 0.77 to 1.40; 5 trials, 864 participants), reoperation (RR 2.92, 95% CI 0.58 to 14.61; 2 trials, 612 participants), or postoperative bile leak (RR 1.00, 95% CI 0.67 to 1.48; 4 trials, 782 participants), as the certainty of evidence was very low for all these outcomes. Fibrin-based haemostatic agents versus non-fibrin-based haemostatic agents Sixteen trials (1944 participants) compared FBHAs with non-FBHAs. All trials had at least one domain at high or unclear risk of bias. Twelve trials were at high risk of bias related to blinding. It is unclear if FBHAs compared with non-FBHAs have an effect on perioperative mortality (RR 1.03, 95% CI 0.62 to 1.72; 11 trials, 1436 participants), postoperative transfusion (RR 0.92, 95% CI 0.68 to 1.25; 7 trials, 599 participants), reoperation (RR 0.48, 95% CI 0.25 to 0.90; 3 trials, 358 participants), or postoperative bile leak (RR 1.15, 95% CI 0.60 to 2.21; 9 trials, 1115 participants), as the certainty of evidence was very low for all these outcomes. FBHAs compared with non-FBHAs may have little or no effect on the risk of serious adverse events (RR 0.99, 95% CI 0.95 to 1.03; 9 trials, 1176 participants; low-certainty evidence).
AUTHORS' CONCLUSIONS
The evidence for the outcomes in both comparisons (FBHAs versus no intervention and FBHAs versus non-FBHAs) was of very low certainty (or low certainty in one instance) and cannot justify the routine use of FBHAs to reduce blood loss in adult liver resection. While the meta-analysis showed a reduced risk of reoperation with FBHAs compared with non-FBHAs, the analysis was confounded by the small number of trials reporting the event and the risk of bias in all these trials. Future trials should focus on the use of FBHAs in people undergoing liver resection who are at particularly high risk of bleeding. Investigators should evaluate clinically meaningful and patient-important outcomes and follow the SPIRIT and CONSORT statements.
Topics: Adult; Humans; Blood Loss, Surgical; Fibrin; Hemostatics; Liver; Quality of Life
PubMed: 37551841
DOI: 10.1002/14651858.CD010872.pub2 -
Inflammation Research : Official... Oct 2017Delta neutrophil index (DNI) representing the number of immature granulocytes is an emerging marker used in diagnosis of infections and prediction of mortality in... (Meta-Analysis)
Meta-Analysis Review
OBJECTIVE
Delta neutrophil index (DNI) representing the number of immature granulocytes is an emerging marker used in diagnosis of infections and prediction of mortality in infected patients. The present study evaluated the diagnostic accuracy of DNI as a predictive and prognostic factor in infected patients.
METHODS
We performed a PubMed search on January 1st, 2017 and identified studies that evaluated DNI as either a predictive or prognostic factor in infected patients. Studies with appropriate information to construct 2 × 2 contingency tables were extracted. We calculated pooled sensitivity and specificity. Meta-analysis of the multivariate logistic regression data set was performed to assess whether DNI functions as an independent factor.
RESULTS
Overall, 12 articles fulfilled the inclusion criteria and a total of 499 cases and 9549 controls were examined. As a predictive factor of infection, DNI's pooled sensitivity was 0.67 (95% CI 0.62-0.71, I = 86.0%) and pooled specificity was 0.94 (95% CI 0.94-0.95, I = 92.8%). Area under the receiver operating characteristics (ROC) curve was 0.89. As a prognostic factor for death in infected patients, DNI's pooled sensitivity was 0.70 (95% CI 0.56-0.81, I = 0.0%) and pooled specificity was 0.78 (95% CI 0.73-0.83, I = 26.6%). Area under the ROC curve was 0.84. Meta-analysis of the multivariate logistic regression data set showed insignificant results.
CONCLUSIONS
DNI is a potentially useful diagnostic tool and predicts mortality among infected patients and should be more widely used in the clinical practice.
Topics: Animals; Biomarkers; Humans; Infections; Neutrophils; Prognosis; Sensitivity and Specificity
PubMed: 28646289
DOI: 10.1007/s00011-017-1066-y -
Frontiers in Medicine 2022The aim of this study was to assess the accuracy of prenatal imaging for the diagnosis of congenital Zika syndrome.
OBJECTIVE
The aim of this study was to assess the accuracy of prenatal imaging for the diagnosis of congenital Zika syndrome.
DATA SOURCES
Medline (via Pubmed), PubMed, Scopus, Web of Science, and Google Scholar from inception to March 2022. Two researchers independently screened study titles and abstracts for eligibility.
STUDY ELIGIBILITY CRITERIA
Observational studies with Zika virus-infected pregnant women were included. The index tests included ultrasound and/or magnetic resonance imaging. The reference standard included (1) Zika infection-related perinatal death, stillbirth, and neonatal death within the first 48 h of birth, (2) neonatal intensive care unit admission, and (3) clinically defined adverse perinatal outcomes.
SYNTHESIS METHODS
We extracted 2 × 2 contingency tables. Pooled sensitivity and specificity were estimated using the random-effects bivariate model and assessed the summary receiver operating characteristic (ROC) curve. Risk of bias was assessed using QUADAS 2 tool. The certainty of the evidence was evaluated with grading of recommendations.
RESULTS
We screened 1,459 references and included 18 studies (2359 pregnant women, 347 fetuses with confirmed Zika virus infection). Twelve studies (67%) were prospective cohorts/case series, and six (37%) were retrospective cohort/case series investigations. Fourteen studies (78%) were performed in endemic regions. Ten studies (56%) used prenatal ultrasound only, six (33%) employed ultrasound and fetal MRI, and two studies (11%) used prenatal ultrasound and postnatal fetal MRI. A total of six studies (ultrasound only) encompassing 780 pregnant women (122 fetuses with confirmed Zika virus infection) reported relevant data for meta-analysis (gestation age at which ultrasound imagining was captured ranged from 16 to 34 weeks). There was large heterogeneity across studies regarding sensitivity (range: 12 to 100%) and specificity (range: 50 to 100%). Under a random-effects model, the summary sensitivity of ultrasound was 82% (95% CI, 19 to 99%), and the summary specificity was 97% (71 to 100%). The area under the ROC curve was 97% (95% CI, 72 to 100%), and the summary diagnostic odds ratio was 140 (95% CI, 3 to 7564, < 0.001). The overall certainty of the evidence was "very low".
CONCLUSION
Ultrasound may be useful in improving the diagnostic accuracy of Zika virus infection in pregnancy. However, the evidence is still substantially uncertain due to the methodological limitations of the available studies. Larger, properly conducted diagnostic accuracy studies of prenatal imaging for the diagnosis of congenital Zika syndrome are warranted.
SYSTEMATIC REVIEW REGISTRATION
Identifier [CRD42020162914].
PubMed: 36250095
DOI: 10.3389/fmed.2022.962765 -
Chinese Medical Journal Aug 2016Periprosthetic joint infection (PJI) is the main cause of failure following total joint arthroplasty. Until now, the diagnosis of PJI is still confronted with technical... (Meta-Analysis)
Meta-Analysis Review
BACKGROUND
Periprosthetic joint infection (PJI) is the main cause of failure following total joint arthroplasty. Until now, the diagnosis of PJI is still confronted with technical limitations, and the question of whether synovial fluid biomarker, C-reactive protein (CRP), can provide high value in the diagnosis of PJI remains unanswered and, therefore, was the aim of the study.
METHODS
First, we conducted a systematic review on CRP in the diagnosis of PJI by searching online databases using keywords such as "periprosthetic joint infection", "synovial fluid", and "C-reactive protein". Eligible studies providing sufficient data to construct 2 × 2 contingency tables were then selected based on the list of criteria and the quality of included studies was assessed subsequently. Finally, the reported sensitivity, specificity, diagnostic odds ratio (DOR), summary receiver operating characteristic (SROC) curve, and the area under the SROC (AUSROC) were pooled together and used to evaluate overall diagnostic performance.
RESULTS
Seven studies were included in our review, six of which comprising a total of 456 participants were further investigated in our meta-analysis. The pooled sensitivity, specificity, and DOR were 0.92 (95% confidence interval [CI]: 0.86-0.96), 0.90 (95% CI: 0.87-0.93), and 101.40 (95% CI: 48.07-213.93), respectively. The AUSROC was 0.9663 (standard error, 0.0113).
CONCLUSIONS
Synovial fluid CRP is a good biomarker for the diagnosis of PJI with high sensitivity and specificity.
Topics: Arthroplasty, Replacement, Hip; Arthroplasty, Replacement, Knee; Biomarkers; C-Reactive Protein; Female; Humans; Male; Prosthesis-Related Infections; Synovial Fluid
PubMed: 27503025
DOI: 10.4103/0366-6999.187857 -
Liver Transplantation : Official... Dec 2017Early detection of small hepatocellular carcinoma (HCC) lesions can improve longterm patient survival. A systematic review and meta-analysis of the diagnostic... (Comparative Study)
Comparative Study Meta-Analysis Review
Early detection of small hepatocellular carcinoma (HCC) lesions can improve longterm patient survival. A systematic review and meta-analysis of the diagnostic performance of gadoxetic acid disodium (Gd-EOB-DTPA)-enhanced magnetic resonance imaging (MRI) and multidetector computed tomography (MDCT) was performed in diagnosing small HCCs measuring up to 2 cm (≤2 cm). Two investigators searched multiple databases for studies in which the performances of either Gd-EOB-DTPA-enhanced MRI or MDCT were reported with sufficient data to construct 2 × 2 contingency tables for diagnosing HCCs up to 2 cm on a per-lesion or per-patient level. Diagnostic performances were quantitatively pooled by a bivariate random-effect model with further meta-regression and subgroup analyses. A total of 27 studies (14 on Gd-EOB-DTPA-enhanced MRI, 9 on MDCT, and 4 on both) were included, enrolling a total of 1735 patients on Gd-EOB-DTPA-enhanced MRI and 1781 patients on MDCT. Gd-EOB-DTPA-enhanced MRI demonstrated significantly higher overall sensitivity than did MDCT (0.96 versus 0.65; P < 0.01), without substantial loss of specificity (0.94 versus 0.98; P > 0.05). Area under the summary receiver operating characteristic curve was 0.97 with Gd-EOB-DTPA-enhanced MRI and 0.85 with MDCT. Regarding Gd-EOB-DTPA-enhanced MRI, sensitivity was significantly higher for studies from non-Asian countries than Asian countries (0.96 versus 0.93; P < 0.01), for retrospective studies than prospective studies (0.95 versus 0.91; P < 0.01), and for those with Gd-EOB-DTPA injection rate ≥ 1.5 mL/s than that of <1.5 mL/s (0.97 versus 0.90; P < 0.01). In conclusion, Gd-EOB-DTPA-enhanced MRI demonstrated higher sensitivity and overall diagnostic accuracy than MDCT, and thus should be the preferred imaging modality for diagnosing small HCCs measuring up to 2 cm. Liver Transplantation 23 1505-1518 2017 AASLD.
Topics: Carcinoma, Hepatocellular; Contrast Media; Early Detection of Cancer; Gadolinium DTPA; Humans; Liver Neoplasms; Magnetic Resonance Imaging; Multidetector Computed Tomography; ROC Curve
PubMed: 28886231
DOI: 10.1002/lt.24867 -
Surgical Endoscopy Apr 2021In the past decade, deep learning has revolutionized medical image processing. This technique may advance laparoscopic surgery. Study objective was to evaluate whether... (Meta-Analysis)
Meta-Analysis Review
BACKGROUND
In the past decade, deep learning has revolutionized medical image processing. This technique may advance laparoscopic surgery. Study objective was to evaluate whether deep learning networks accurately analyze videos of laparoscopic procedures.
METHODS
Medline, Embase, IEEE Xplore, and the Web of science databases were searched from January 2012 to May 5, 2020. Selected studies tested a deep learning model, specifically convolutional neural networks, for video analysis of laparoscopic surgery. Study characteristics including the dataset source, type of operation, number of videos, and prediction application were compared. A random effects model was used for estimating pooled sensitivity and specificity of the computer algorithms. Summary receiver operating characteristic curves were calculated by the bivariate model of Reitsma.
RESULTS
Thirty-two out of 508 studies identified met inclusion criteria. Applications included instrument recognition and detection (45%), phase recognition (20%), anatomy recognition and detection (15%), action recognition (13%), surgery time prediction (5%), and gauze recognition (3%). The most common tested procedures were cholecystectomy (51%) and gynecological-mainly hysterectomy and myomectomy (26%). A total of 3004 videos were analyzed. Publications in clinical journals increased in 2020 compared to bio-computational ones. Four studies provided enough data to construct 8 contingency tables, enabling calculation of test accuracy with a pooled sensitivity of 0.93 (95% CI 0.85-0.97) and specificity of 0.96 (95% CI 0.84-0.99). Yet, the majority of papers had a high risk of bias.
CONCLUSIONS
Deep learning research holds potential in laparoscopic surgery, but is limited in methodologies. Clinicians may advance AI in surgery, specifically by offering standardized visual databases and reporting.
Topics: Deep Learning; Diagnostic Tests, Routine; Female; Humans; Laparoscopy; Male
PubMed: 33398560
DOI: 10.1007/s00464-020-08168-1