-
BMJ Open Aug 2023Childhood cataract is a chronic condition that may interfere with the child's learning capacities. We aimed to investigate whether childhood cataract influences academic...
OBJECTIVES
Childhood cataract is a chronic condition that may interfere with the child's learning capacities. We aimed to investigate whether childhood cataract influences academic development by comparing school performance in reading and mathematics in children with cataract to a matched control group.
DESIGN
Nationwide registry-based cohort study.
SETTINGS
Two surgical centres that perform all treatments for childhood cataract in Denmark.
PARTICIPANTS
Children born between 2000 and 2009 diagnosed with cataract before 10 years of age (n=275) and an age-matched and sex-matched control group (n=2473).
MAIN OUTCOME MEASURES
School performance was assessed as test scores in national tests performed at regular intervals from grade 2 to grade 8 in reading and mathematics. Analyses were corrected for birth origin, child somatic and mental disorder and parental socioeconomic status and mental disorders.
RESULTS
Of 275 children, 85 (30.9%) were operated for bilateral cataract, 79 (28.7%) unilateral cataract and 111 (40,4%) were not operated. We found that children with cataract have lower participation rate in the tests (62.5%) compared with the control cohort (77.2%) (p value=0.0001). After adjusting the pooled analyses for birth origin, somatic and mental disease in the child and parental socioeconomic status and mental disorders, we found that the children with cataract scored significantly lower in mathematics compared with those without cataract (mean difference=-4.78, 95% CI: -8.18 to -1.38, p value=0.006), whereas no difference was found regarding scores in reading (p=0.576). The lower score in mathematics was driven by children who had been operated for bilateral cataract (p-value=0.004).
CONCLUSION
Children with cataract without somatic or neurodevelopmental comorbidities or psychosocial adversities seem to do well in school, whereas children operated for bilateral cataract have higher frequencies of difficulties in mathematical tasks.
Topics: Humans; Child; Cohort Studies; Academic Performance; Cataract; Schools; Comorbidity
PubMed: 37532485
DOI: 10.1136/bmjopen-2023-072984 -
Medical Image Analysis Oct 2023Training segmentation models for medical images continues to be challenging due to the limited availability of data annotations. Segment Anything Model (SAM) is a...
Training segmentation models for medical images continues to be challenging due to the limited availability of data annotations. Segment Anything Model (SAM) is a foundation model trained on over 1 billion annotations, predominantly for natural images, that is intended to segment user-defined objects of interest in an interactive manner. While the model performance on natural images is impressive, medical image domains pose their own set of challenges. Here, we perform an extensive evaluation of SAM's ability to segment medical images on a collection of 19 medical imaging datasets from various modalities and anatomies. In our experiments, we generated point and box prompts for SAM using a standard method that simulates interactive segmentation. We report the following findings: (1) SAM's performance based on single prompts highly varies depending on the dataset and the task, from IoU=0.1135 for spine MRI to IoU=0.8650 for hip X-ray. (2) Segmentation performance appears to be better for well-circumscribed objects with prompts with less ambiguity such as the segmentation of organs in computed tomography and poorer in various other scenarios such as the segmentation of brain tumors. (3) SAM performs notably better with box prompts than with point prompts. (4) SAM outperforms similar methods RITM, SimpleClick, and FocalClick in almost all single-point prompt settings. (5) When multiple-point prompts are provided iteratively, SAM's performance generally improves only slightly while other methods' performance improves to the level that surpasses SAM's point-based performance. We also provide several illustrations for SAM's performance on all tested datasets, iterative segmentation, and SAM's behavior given prompt ambiguity. We conclude that SAM shows impressive zero-shot segmentation performance for certain medical imaging datasets, but moderate to poor performance for others. SAM has the potential to make a significant impact in automated medical image segmentation in medical imaging, but appropriate care needs to be applied when using it. Code for evaluation SAM is made publicly available at https://github.com/mazurowski-lab/segment-anything-medical-evaluation.
Topics: Humans; Brain Neoplasms; S-Adenosylmethionine; Tomography, X-Ray Computed
PubMed: 37595404
DOI: 10.1016/j.media.2023.102918 -
International Journal of Molecular... Feb 2024Immunoassays (IAs) with fluorescence-based detection are already well-established commercialized biosensing methods, such as enzyme-linked immunosorbent assay (ELISA)... (Review)
Review
Immunoassays (IAs) with fluorescence-based detection are already well-established commercialized biosensing methods, such as enzyme-linked immunosorbent assay (ELISA) and lateral flow immunoassay (LFIA). Immunoassays with surface-enhanced Raman spectroscopy (SERS) detection have received significant attention from the research community for at least two decades, but so far they still lack a wide clinical commercial application. This review, unlike any other review that we have seen, performs a three-dimensional performance comparison of SERS IAs vs. fluorescence IAs. First, we compared the limit of detection (LOD) as a key performance parameter for 30 fluorescence and 30 SERS-based immunoassays reported in the literature. We also compared the clinical performances of a smaller number of available reports for SERS vs. fluorescence immunoassays (FIAs). We found that the median and geometric average LODs are about 1.5-2 orders of magnitude lower for SERS-based immunoassays in comparison to fluorescence-based immunoassays. For instance, the median LOD for SERS IA is 4.3 × 10 M, whereas for FIA, it is 1.5 × 10 M. However, there is no significant difference in average relative standard deviation (RSD)-both are about 5-6%. The analysis of sensitivity, selectivity, and accuracy reported for a limited number of the published clinical studies with SERS IA and FIA demonstrates an advantage of SERS IA over FIA, at least in terms of the median value for all three of those parameters. We discussed common and specific challenges to the performances of both SERS IA and FIA, while proposing some solutions to mitigate those challenges for both techniques. These challenges include non-specific protein binding, non-specific interactions in the immunoassays, sometimes insufficient reproducibility, relatively long assay times, photobleaching, etc. Overall, this review may be useful for a large number of researchers who would like to use immunoassays, but particularly for those who would like to make improvements and move forward in both SERS-based IAs and fluorescence-based IAs.
Topics: Reproducibility of Results; Spectrum Analysis, Raman; Immunoassay; Coloring Agents; Enzyme-Linked Immunosorbent Assay; Gold; Metal Nanoparticles
PubMed: 38396756
DOI: 10.3390/ijms25042080 -
Journal of Reconstructive Microsurgery Nov 2023Preparation of the recipient vessels is a crucial step in autologous breast reconstruction, with limited opportunity for resident training intraoperatively. The...
BACKGROUND
Preparation of the recipient vessels is a crucial step in autologous breast reconstruction, with limited opportunity for resident training intraoperatively. The Blue-Blood-infused porcine chest wall-a cadaveric pig thorax embedded in a mannequin shell, connected to a saline perfusion system-is a novel, cost-effective ($55) simulator of internal mammary artery (IMA) dissection and anastomosis intended to improve resident's comfort, safety, and expertise with all steps of this procedure. The purpose of this study was to assess the effect of the use of this chest wall model on resident's confidence in performing dissection and anastomosis of the IMA, as well as obtain resident's and faculty's perspectives on model realism and utility.
METHODS
Plastic surgery residents and microsurgery faculty at the University of Wisconsin were invited to participate. One expert microsurgeon led individual training sessions and performed as the microsurgical assistant. Participants anonymously completed surveys prior to and immediately following their training session to assess their change in confidence performing the procedure, as well as their perception of model realism and utility as a formal microsurgical training tool on a five-point scale.
RESULTS
Every participant saw improvement in confidence after their training session in a minimum of one of seven key procedural steps identified. Of participants who had experience with this procedure in humans, the majority rated model anatomy and performance of key procedural steps as "very" or "extremely" realistic as compared with humans. 100% of participants believed practice with this model would improve residents' ability to perform this operation in the operating room and 100% of participants would recommend this model be incorporated into the microsurgical training curriculum.
CONCLUSION
The Blue-Blood porcine chest wall simulator increases trainee confidence in performing key steps of IMA dissection and anastomosis and is perceived as valuable to residents and faculty alike.
Topics: Humans; Swine; Animals; Internship and Residency; Clinical Competence; Education, Medical, Graduate; Simulation Training; Thorax
PubMed: 36931312
DOI: 10.1055/a-2057-0766 -
Neurourology and Urodynamics Sep 2023Practice patterns around the use of urodynamic evaluation (UDS) for benign prostatic hyperplasia (BPH) surgery are largely undefined. As such, we investigated factors...
INTRODUCTION
Practice patterns around the use of urodynamic evaluation (UDS) for benign prostatic hyperplasia (BPH) surgery are largely undefined. As such, we investigated factors associated with the use of UDS for BPH.
METHODS
We used American Board of Urology case log data from 2008 to 2020, to compare patient- and surgeon-sided factors associated with UDS utilization and BPH surgeries. We performed logistic regression models to identify factors independently associated with UDS usage for BPH.
RESULTS
Among urologists performing UDS, the majority (80%) self-identified as general urologists and practiced in a private practice group (69%). Compared with urologists who performed no UDS, urologists who performed any UDS for BPH were more likely to be from the Mid-Atlantic (20.3% vs. 10.6%, p < 0.01) and practice in regions with populations of >1 000 000 (34.7% vs. 28.5%, p < 0.01). Overall, UDS utilization declined over time (odds ratio [OR]: 0.95 year-to-year, 95% confidence interval [CI]: 0.91-0.99). In adjusted analyses, the odds of performing UDS was higher among male (OR: 2.19, 95% CI: 1.17-4.09), older (OR: 1.05, 95% CI: 1.03-1.06), and female pelvic medicine and reconstructive surgery subspecialty (OR: 3.23, 95% CI: 2.01-5.2) urologists. Additionally, performing UDS for BPH was associated with higher BPH surgical case volume (OR: 1.004, 95% CI: 1.001-1.008).
CONCLUSION
There is a significant practice variation in use of UDS for BPH. Although overall BPH surgeries are increasing, urologists are increasingly less likely to perform UDS for BPH. Specifically, urologists who perform UDS have significantly higher BPH case volume than those who do not perform UDS, suggesting that UDS usage may not factor into BPH surgery decision-making.
Topics: Humans; Male; Female; Prostatic Hyperplasia; Urodynamics; Practice Patterns, Physicians'; Urology; Urologists
PubMed: 37395472
DOI: 10.1002/nau.25239 -
Surgical Laparoscopy, Endoscopy &... Apr 2024As the use of the robotic platform increases among general surgeons, the amount of robotic cholecystectomies is expected to increase as well. The use of intraoperative...
BACKGROUND
As the use of the robotic platform increases among general surgeons, the amount of robotic cholecystectomies is expected to increase as well. The use of intraoperative cholangiography is valuable in assessing for choledocholithiasis. We describe our technique of performing robotic intraoperative cholangiograms with choledochoscopy. Out technique aids in efficiency since no undocking is required.
METHODS
Preoperatively, the decision to perform a cholangiogram is made based on physical exam, labs, and imaging findings. The procedure begins with obtaining a critical view of safety. The robotic arms are positioned in a manner that allows all 4 robotic arms to remain docked. A ductotomy is made and the cholangiocatheter is introduced. The cholangiogram images are then interpreted and if a stone is seen in the common bile duct we will then perform a transcystic common bile duct exploration using the SpyGlass Discover digital. A complete cholangiogram is then performed. The cystic duct is secured and the gallbladder is removed from the liver bed. The patients are watched overnight and discharged on postoperative day 1.
CONCLUSIONS
A robotic approach to performing a transcystic common bile duct exploration is a safe and reproducible treatment method for choledocholithiasis. Our approach offers an advantage since no undocking is required.
Topics: Humans; Choledocholithiasis; Gallbladder; Robotic Surgical Procedures; Laparoscopy; Common Bile Duct; Cholangiography; Cholecystectomy, Laparoscopic
PubMed: 38359352
DOI: 10.1097/SLE.0000000000001258 -
Frontiers in Robotics and AI 2023Nowadays, robotics applications requiring the execution of complex tasks in real-world scenarios are still facing many challenges related to highly unstructured and...
Nowadays, robotics applications requiring the execution of complex tasks in real-world scenarios are still facing many challenges related to highly unstructured and dynamic environments in domains such as emergency response and search and rescue where robots have to operate for prolonged periods trading off computational performance with increased power autonomy and . In particular, there is a crucial need for robots capable of adapting to such settings while at the same time providing robustness and extended power autonomy. A possible approach to overcome the conflicting demand of a computational performing system with the need for long power autonomy is represented by cloud robotics, which can boost the computational capabilities of the robot while reducing the energy consumption by exploiting the offload of resources to the cloud. Nevertheless, the communication constraint due to limited bandwidth, latency, and connectivity, typical of field robotics, makes cloud-enabled robotics solutions challenging to deploy in real-world applications. In this context, we designed and realized the XBot2D software architecture, which provides a hybrid cloud manager capable of dynamically and seamlessly allocating robotics skills to perform a distributed computation based on the current network condition and the required latency, and computational/energy resources of the robot in use. The proposed framework leverage on the two dimensions, i.e., 2D (local and cloud), in a transparent way for the user, providing support for Real-Time (RT) skills execution on the local robot, as well as machine learning and A.I. resources on the cloud with the possibility to automatically relocate the above based on the required performances and communication quality. XBot2D implementation and its functionalities are presented and validated in realistic tasks involving the CENTAURO robot and the Amazon Web Service Elastic Computing Cloud (AWS EC2) infrastructure with different network conditions.
PubMed: 37860633
DOI: 10.3389/frobt.2023.1168694 -
BMC Health Services Research Jan 2024No-show to medical appointments has significant adverse effects on healthcare systems and their clients. Using machine learning to predict no-shows allows managers to...
BACKGROUND
No-show to medical appointments has significant adverse effects on healthcare systems and their clients. Using machine learning to predict no-shows allows managers to implement strategies such as overbooking and reminders targeting patients most likely to miss appointments, optimizing the use of resources.
METHODS
In this study, we proposed a detailed analytical framework for predicting no-shows while addressing imbalanced datasets. The framework includes a novel use of z-fold cross-validation performed twice during the modeling process to improve model robustness and generalization. We also introduce Symbolic Regression (SR) as a classification algorithm and Instance Hardness Threshold (IHT) as a resampling technique and compared their performance with that of other classification algorithms, such as K-Nearest Neighbors (KNN) and Support Vector Machine (SVM), and resampling techniques, such as Random under Sampling (RUS), Synthetic Minority Oversampling Technique (SMOTE) and NearMiss-1. We validated the framework using two attendance datasets from Brazilian hospitals with no-show rates of 6.65% and 19.03%.
RESULTS
From the academic perspective, our study is the first to propose using SR and IHT to predict the no-show of patients. Our findings indicate that SR and IHT presented superior performances compared to other techniques, particularly IHT, which excelled when combined with all classification algorithms and led to low variability in performance metrics results. Our results also outperformed sensitivity outcomes reported in the literature, with values above 0.94 for both datasets.
CONCLUSION
This is the first study to use SR and IHT methods to predict patient no-shows and the first to propose performing z-fold cross-validation twice. Our study highlights the importance of avoiding relying on few validation runs for imbalanced datasets as it may lead to biased results and inadequate analysis of the generalization and stability of the models obtained during the training stage.
Topics: Humans; Algorithms; Benchmarking; Brazil; Machine Learning; Decision Support Techniques
PubMed: 38183029
DOI: 10.1186/s12913-023-10418-6 -
The Laryngoscope Mar 2024To evaluate the literature and summarize cochlear implantation (CI) outcomes after intralabyrinthine schwannoma (ILS) excision and tumor observation with CI. (Review)
Review
OBJECTIVE
To evaluate the literature and summarize cochlear implantation (CI) outcomes after intralabyrinthine schwannoma (ILS) excision and tumor observation with CI.
DATA SOURCES
OVID Medline, Embase, Web of Science; conception to 2024.
REVIEW METHODS
A literature review was performed using subject headings, MeSH terms, and keywords. Abstracts and full texts were reviewed by two independent reviewers and adjudicated by a third. Inclusion criteria included studies with ILS and CI with reported audiologic outcomes. Subjects were analyzed into two groups, ILS resection with CI and in situ ILS with CI. Patients with NF2 were included. The main outcome of interest was CI audiometric performance level, with secondary outcomes of CI user status and open-set speech attainment.
RESULTS
There were 29 articles with a total of 93 patients who met inclusion criteria. The resection group had 17% low performers, 44% intermediate performers, and 38% high performers. The in situ group had 40% low performers, 32% intermediate performers, 27% high performers. The resection group had 69 patients with 97% maintaining user status and 92% with open-set speech recognition. The observation group had 24 patients, with 87% user rate and 86% achieving open-set speech recognition. There was a greater percentage of NF2 diagnosis seen in the in situ group.
CONCLUSION
There is a paucity of literature on CI and ILS. Patients are managed with both resection of tumor and implantation in situ. Early data are limited, with improvement in hearing outcomes and high user rates in both populations.
LEVEL OF EVIDENCE
N/A Laryngoscope, 2024.
PubMed: 38554009
DOI: 10.1002/lary.31422 -
Journal of Alzheimer's Disease : JAD 2024A screening tool sensitive to Alzheimer's disease (AD) risk factors, such as amyloid-β (Aβ) deposition, and subtle cognitive changes, best elicited by complex everyday...
BACKGROUND
A screening tool sensitive to Alzheimer's disease (AD) risk factors, such as amyloid-β (Aβ) deposition, and subtle cognitive changes, best elicited by complex everyday tasks, is needed.
OBJECTIVE
To determine if grocery shopping performance could differentiate older adults at elevated risk of developing AD (OAer), older adults at low risk of developing AD (OAlr), and young adults (YA), and if amount of Aβ deposition could predict grocery shopping performance in older adults (OA).
METHODS
Twenty-one OAer (78±5 years), 33 OAlr (78±5 years), and 28 YA (31±3 years) performed four grocery shopping trials, with the best and worst performances analyzed. Measures included trial time, number of correct items, number of grocery note fixations, and number of fixations and percentage of time fixating on the correct shelving unit, correct brand, and correct shelf. Linear mixed effects models compared measures by performance rank (best, worst) and group (OAer, OAlr, YA), and estimated the effect of Aβ deposition on measures in OA.
RESULTS
Relative to their best performance, OAer and OAlr exhibited more correct shelving unit fixations and correct brand fixations during their worst performance, while YA did not. Within OA's worst performance, greater Aβ deposition was associated with a smaller percentage of time fixating on the correct shelving unit, correct shelf, and correct brand. Within OA, greater Aβ deposition was associated with more grocery note fixations.
CONCLUSIONS
OA with elevated Aβ deposition may exhibit subtle working memory impairments and less efficient visual search strategies while performing a cognitively demanding everyday task.
Topics: Humans; Aged; Male; Female; Amyloid beta-Peptides; Adult; Aged, 80 and over; Neuropsychological Tests; Positron-Emission Tomography; Alzheimer Disease; Young Adult; Aging; Activities of Daily Living; Brain
PubMed: 38820016
DOI: 10.3233/JAD-231108