An astonishing regularity in student learning rate.

Leveraging a scientific infrastructure for exploring how students learn, we have developed cognitive and statistical models of skill acquisition and used them to understand fundamental similarities and differences across learners. Our primary question was why do some students learn faster than others? Or, do they? We model data from student performance on groups of tasks that assess the same skill component and that provide follow-up instruction on student errors. Our models estimate, for both students and skills, initial correctness and learning rate, that is, the increase in correctness after each practice opportunity. We applied our models to 1.3 million observations across 27 datasets of student interactions with online practice systems in the context of elementary to college courses in math, science, and language. Despite the availability of up-front verbal instruction, like lectures and readings, students demonstrate modest initial prepractice performance, at about 65% accuracy. Despite being in the same course, students' initial performance varies substantially from about 55% correct for those in the lower half to 75% for those in the upper half. In contrast, and much to our surprise, we found students to be astonishingly similar in estimated learning rate, typically increasing by about 0.1 log odds or 2.5% in accuracy per opportunity. These findings pose a challenge for theories of learning to explain the odd combination of large variation in student initial performance and striking regularity in student learning rate.

Point-of-Care Ultrasound Psychomotor Learning Curves: A Systematic Review of the Literature.

OBJECTIVES: Use of point-of-care ultrasound (POCUS) in clinical medicine and inclusion in medical training is increasing. Some professional societies recommend that 25-50 POCUS examinations be completed for each application learned; however, the amount of practice required is not well studied. As such, a better understanding of the learning curves of POCUS psychomotor skills is needed. This systematic review characterizes the learning curves for POCUS psychomotor skill acquisition. METHODS: With the assistance of a research librarian, the available literature through August 28, 2023, was identified. The titles and abstracts, and then the full text were reviewed by two reviewers to screen for inclusion. All studies included after full-text review then underwent data extraction and analysis. RESULTS: The search identified 893 unique studies. Forty-five studies underwent full-text review, with 17 meeting full inclusion criteria. Substantial heterogeneity was noted in study design, duration of education, number and type of learners, and methods for statistical analysis. Clear and validated definitions for learning endpoints, such as plateau points or competency, are lacking. Learning curves and endpoints differ for different applications of POCUS. CONCLUSION: The results are overall supportive of the recommendations to complete 25-50 examinations per application of POCUS learned. However, specific applications require more practice than others. Certain applications, such as cardiac and the Focused Assessment with Sonography in Trauma (FAST) exams, are closer to 50; while others, such as soft tissue, airway, and eye require no more than 25.

The learning curve for minimally invasive Achilles repair using the "lumbar puncture needle and oval forceps" technique.

INTRODUCTION: An acute Achilles tendon rupture represents a common tendon injury, and its operative methods have been developed over the years. This study aimed to quantify the learning curve for the minimally invasive acute Achilles tendon rupture repair. METHODS: From May 2020 to June 2022, sixty-seven patient cases who received minimally invasive tendon repair were reviewed. Baseline data and operative details were collected. The cumulative summation (CUSUM) control chart was used for the learning curve analyses. Achilles tendon rupture score (ATRS), American Orthopedic Foot and Ankle Society (AOFAS) ankle/hindfoot score, and visual analog scale (VAS) at 3/6/9/12 months were calculated to assess the clinical outcomes. RESULTS: Thirty-six cases underwent at least a year of follow up and were enrolled in this study. The gender ratio and average age were 80.5% and 32.5 years. The linear equation fitted well (R2 = 0.95), and CUSUM for operative time peaked in the 12th case, which was divided into the learning phase (n = 12) and master phase (n = 24). No significant difference was detected between the two groups in clinical variables, except for the operative time (71.1 ± 13.2 min vs 45.8 ± 7.2 min, p = 0.004). Moreover, we detected one case with a suture reaction and treated it properly. CONCLUSION: Minimally invasive Achilles repair provides an opportunity for early rehabilitation. Notably, the learning curve showed that the "lumbar puncture needle and oval forceps" technique was accessible to surgeons.

Learning curves for robotic-assisted spine surgery: an analysis of the time taken for screw insertion, robot setting, registration, and fluoroscopy.

PURPOSE: The purpose of this study was to clarify the learning curve for robotic-assisted spine surgery. We analyzed the workflow in robotic-assisted spine surgery and investigated how much experience is required to become proficient in robotic-assisted spine surgery. METHODS: The data were obtained from consecutive 125 patients who underwent robotic-assisted screw placement soon after introducing a spine robotic system at a single center from April 2021 to January 2023. The 125 cases were divided into phases 1-5 of sequential groups of 25 cases each and compared for screw insertion time, robot setting time, registration time, and fluoroscopy time. RESULTS: There were no significant differences in age, body mass index, intraoperative blood loss, number of fused segments, operation time, or operation time per segment between the 5 phases. There were significant differences in screw insertion time, robot setting time, registration time, and fluoroscopy time between the 5 phases. The screw insertion time, robot setting time, registration time, and fluoroscopy time in phase 1 were significantly longer than those in phases 2, 3, 4, and 5. CONCLUSION: In an analysis of 125 cases after the introduction of the spine robotic system, the screw insertion time, robot setting time, registration time, and fluoroscopy time were significantly longer in the 25 cases in the period initially after introduction. The times were not significantly different in the subsequent 100 cases. Surgeons can be proficient in robotic-assisted spine surgery after their experience with 25 cases.

The Bigger Fish: A Comparison of Meta-Learning QSAR Models on Low-Resourced Aquatic Toxicity Regression Tasks.

Toxicological information as needed for risk assessments of chemical compounds is often sparse. Unfortunately, gathering new toxicological information experimentally often involves animal testing. Simulated alternatives, e.g., quantitative structure-activity relationship (QSAR) models, are preferred to infer the toxicity of new compounds. Aquatic toxicity data collections consist of many related tasks─each predicting the toxicity of new compounds on a given species. Since many of these tasks are inherently low-resource, i.e., involve few associated compounds, this is challenging. Meta-learning is a subfield of artificial intelligence that can lead to more accurate models by enabling the utilization of information across tasks. In our work, we benchmark various state-of-the-art meta-learning techniques for building QSAR models, focusing on knowledge sharing between species. Specifically, we employ and compare transformational machine learning, model-agnostic meta-learning, fine-tuning, and multi-task models. Our experiments show that established knowledge-sharing techniques outperform single-task approaches. We recommend the use of multi-task random forest models for aquatic toxicity modeling, which matched or exceeded the performance of other approaches and robustly produced good results in the low-resource settings we studied. This model functions on a species level, predicting toxicity for multiple species across various phyla, with flexible exposure duration and on a large chemical applicability domain.

Real-time fMRI neurofeedback compared to cognitive behavioral therapy in a pilot study for the treatment of mild and moderate depression.

Real-time functional magnetic resonance imaging (rt-fMRI) neurofeedback was found to reduce depressive symptoms. However, no direct comparison of drug-free patients with an active psychotherapy control group is available. The present study compared rt-fMRI neurofeedback with cognitive behavioral therapy, as the standard treatment in patients declining anti-depressants. Twenty adult, drug-free patients with mild or moderate depression were non-randomly assigned either to a course of eight half-hour sessions of neurofeedback targeting the left medial prefrontal cortex (N = 12) or to a 16-session course of cognitive behavioral therapy (N = 8). Montgomery-Asberg Depression Rating Scale was introduced at baseline, mid-treatment, and end-treatment points. In each group, 8 patients each remained in the study to a mid-treatment evaluation and 6 patients each to the study end-point. ANOVA revealed a depression reduction with a significant effect of Time (F(3,6) = 19.0, p < 0.001, η2 = 0.76). A trend to greater improvement in the cognitive behavioral therapy group compared to neurofeedback emerged (Group × Time; p = 0.078). Percent signal change in the region of interest between up- and down-regulation conditions was significantly correlated with session number (Pearson's r = 0.85, p < 0.001) indicating a learning effect. As limitations, small sample size could lead to insufficient power and non-random allocation to selection bias. Both neurofeedback and cognitive behavioral therapy improved mild and moderate depression. Neurofeedback was not superior to cognitive behavioral therapy. Noteworthy, the neurofeedback training course was associated with continuous improvement in the self-regulation skill, without plateau. This study delivers data to plan clinical trials comparing neurofeedback with cognitive behavioral interventions.

Learning curves of robotic extended totally extraperitoneal (eTEP) hernia repair among two surgeons at a high-volume community hospital: a cumulative sum analysis.

INTRODUCTION: Robotic extended totally extraperitoneal hernia (eTEP) repair is a novel technique for minimally invasive ventral hernia repair with retromuscular placement of mesh. This study aimed to evaluate the learning curve for robotic eTEP hernia repair using risk-adjusted cumulative sum (RA-CUSUM) analysis for two general surgeons-one with dedicated fellowship training in robotic eTEP technique (surgeon 2) and another without robotic eTEP-specific training (surgeon 1). METHODS: We conducted a retrospective analysis of 98 patients undergoing robotic eTEP hernia repair from July 2020 to February 2022 for two surgeons. RA-CUSUM method was applied to the overall operative time (OT) in minutes, adjusting for transversus abdominis release (TAR). RESULTS: Figures 3 (surgeon 1) and 4 (surgeon 2) illustrate the three phases in the RA-CUSUM graphs of OT. For surgeon 1, the cases for each phase were determined: phase 1 (1 to 12), phase 2 (13 to 24), and phase 3 (25 to 51). For surgeon 2, the three phases were similarly determined as 1 to 8, 9 to 32, and 33 to 47, respectively. A significant (p = 0.017) difference existed for the OTs between phases 1 (262 ± 69) and 3 (192 ± 63.0) for surgeon 1. OT compared to the risk-adjusted value stabilized after case 12 and decreased after case 24 for surgeon 1; it began to decrease after case 8 for surgeon 2. CONCLUSIONS: The initial learning curve for surgeon 1 reached its plateau after 12 cases, shorter than comparable studies. This was likely due to the surgeon's intentional focus on learning this technique through courses, proctoring, and active mentorship. The flat learning curve seen in surgeon 2's series illustrates the value of experience gained during fellowship training. Our data support that, given the right resources and support, a short learning curve for eTEP is attainable for community surgeons without prior training in the technique.

Inguinal hernia surgery learning curves by associate clinicians.

BACKGROUND: Open inguinal hernia repair is the most commonly performed procedure in general surgery in sub-Saharan Africa, but data on its learning curve are lacking. This study evaluated the learning curve characteristics to improve surgical training and enable scaling up hernia surgery in low- and middle-income countries. METHODS: Logbook data of associate clinicians enrolled in a surgical training program in Sierra Leone were collected and their first 55 hernia surgeries following the Bassini technique (herniorrhaphies) were analyzed in cohorts of five cases. Studied variables were gradient of decline of operating time, variation in operating time, and length of stay (LOS). Eleven subsequent cohorts of each five herniorrhaphies were investigated. RESULTS: Seventy-five trainees enrolled in the training program between 2011 and 2020 were eligible for inclusion. Thirty-one (41.3%) performed the minimum of 55 herniorrhaphies, and had also complete personal logbook data. Mean operating times dropped from 79.6 (95% CI 75.3-84.0) to 48.6 (95% CI 44.3-52.9) minutes between the first and last cohort, while standard deviation in operating time nearly halved to 15.4 (95% CI 11.7-20.0) minutes, and LOS was shortened by 3 days (8.5 days, 95%CI 6.1-10.8 vs. 5.4 days, 95% 3.1-7.6). Operating times flattened after 31-35 cases which corresponded with 1.5 years of training. CONCLUSIONS: The learning curve of inguinal hernia surgery for associate clinicians flattens after 31-35 procedures. Training programs can be tailored based on this finding. The recorded learning curve may serve as a baseline for future training techniques.

Residency Surgical BPH Training Paradigms from MIST to HOLEP.

PURPOSE OF REVIEW: Benign prostatic hyperplasia (BPH) is a common medical condition of older men that often requires medical or surgical therapy. Surgical options for BPH have grown exponentially over the last two decades. The numerous options and/or lack of access to them can make it challenging for new trainees to gain proficiency. We examine the literature for available BPH surgical simulators, learning curves, and training pathways. RECENT FINDINGS: Each BPH surgical therapy has a learning curve which must be overcome. There is an abundance of TURP simulators which have shown face, content, and construct validity in the literature. Similarly, laser therapies have validated simulators. Newer technologies do have available simulators, but they have not been validated. There are strategies to improve learning and outcomes, such as having a structured training program. Simulators are available for BPH surgical procedures and some have been implemented in urology residencies. It is likely that such simulation may make urologists more facile on their learning curves for newer technologies. Further studies are needed. Future directions may include integration of simulator technology into training pathways that include surgical observation and proctorship.

Learning curve for adoption of robot-assisted minimally invasive esophagectomy: a systematic review of oncological, clinical, and efficiency outcomes.

BACKGROUND: Robot-assisted minimally invasive esophagectomy (RAMIE) is gaining increasing popularity as an operative approach. Learning curves to achieve surgical competency in robotic-assisted techniques have shown significant variation in learning curve lengths and outcomes. This study aimed to summarize the current literature on learning curves for RAMIE. METHODS: A systematic review was conducted in line with PRISMA guidelines. Electronic databases PubMed, MEDLINE, and Cochrane Library were searched, and articles reporting on learning curves in RAMIE were identified and scrutinized. Studies were eligible if they reported changes in operative outcomes over time, or learning curves, for surgeons newly adopting RAMIE. RESULTS: Fifteen studies reporting on 1767 patients were included. Nine studies reported on surgeons with prior experience of robot-assisted surgery prior to adopting RAMIE, with only four studies outlining a specified RAMIE adoption pathway. Learning curves were most commonly analyzed using cumulative sum control chart (CUSUM) and were typically reported for lymph node yields and operative times, with significant variation in learning curve lengths (18-73 cases and 20-80 cases, respectively). Most studies reported adoption without significant impact on clinical outcomes such as anastomotic leak; significant learning curves were more likely in studies, which did not report a formal learning or adoption pathway. CONCLUSION: Reported RAMIE adoption phases are variable, with some authors suggesting significant impact to patients. With robust training through formal programmes or proctorship, however, others report RAMIE adoption without impact on clinical outcomes. A formalized adoption curriculum appears critical to prevent adverse effects on operative efficiency and patient care.

Characterizing individual trainee learning curves in surgical training: Challenges and opportunities.

The surgical learning curve is an observable and measurable phenomenon. In the era of competency-based approaches to surgical training, monitoring the trajectory of individual trainee competence attainment could represent a meaningful method of formative and summative assessment. While technology can assist this approach, a number of significant barriers to the implementation of such assessment methods remain, including: accurate data collection, standard setting, and reliable assessment. Translating individual learning curve data into quantifiable case minimum targets in training poses further difficulties, and may not be possible for all procedures, particularly those that are less frequently performed and assessed. In spite of these challenges, significant benefits could be realized through an individualized approach to competency assessment using trainee learning curve data. Tracking competence acquisition against criterion-referenced standards could allow for targeted training and remediation, conforming with modern theories of adult education and empowering trainees to take control of their own learning. Learning curve data could also be used to assess the effects of educational interventions such as simulation-based training on subsequent competence acquisition rates. Ultimately, the individual learning curves of trainees could be used to inform personalised decisions regarding entrustment, credentialing, and certification, allowing training programmes to move beyond minimum operative experience targets as a crude proxy measure of competence.

Determining the frequency of burn wound dressing for clinically competent nursing students: establishing standards based on learning curves.

BACKGROUND: The primary objective of clinical practice in nursing education is to achieve mastery of clinical skills through repetitive practice. Therefore, there exists a correlation between the frequency of skill demonstration and clinical competency. This study aimed to address the following question: How many times should a nursing student perform burn wound dressing to attain clinical competency? METHODS: This time series study was conducted on 41 junior nursing students who were selected through a census sampling method at Neyshabur School of Nursing from spring 2015 to summer 2021. The data collection tool was a researcher-made competency evaluation checklist (CEV). The competency score of each student for each skill demonstraion episode was evaluated using the Competency Evaluation Checklist (CEV) and recorded on the learning curve until a plateau was reached. SPSS16 and repeated measures analysis of variance (ANOVA) were used. RESULTS: The mean competency score of burn wound dressing was 67.5 ± 11.0 (out of 100) during the first attempt, which increased to 95.9 ± 4.3 by the ninth attempt. The learning curve for the burn wound dressing skill reached a plateau after five attempts. Additionally, by the fifth attempt of dressing the burn wound, all students could accurately identify the depth, extent, and severity of the wound. CONCLUSIONS: Establishing a standardized frequency for clinical skill performance and ensuring ample practice opportunities can have significant benefits in nursing education systems. These benefits include long-lasting learning, reduced costs, and improved effectiveness. As a result, nurse managers and lecturers should consider the resources available in their educational systems and strive to provide students with adequate opportunities and a supportive environment to practice their skills.

Cost of simulation-based mastery learning for abdominal ultrasound.

BACKGROUND: Ultrasound is an essential diagnostic examination used in several medical specialties. However, the quality of ultrasound examinations is dependent on mastery of certain skills, which may be difficult and costly to attain in the clinical setting. This study aimed to explore mastery learning for trainees practicing general abdominal ultrasound using a virtual reality simulator and to evaluate the associated cost per student achieving the mastery learning level. METHODS: Trainees were instructed to train on a virtual reality ultrasound simulator until the attainment of a mastery learning level was established in a previous study. Automated simulator scores were used to track performances during each round of training, and these scores were recorded to determine learning curves. Finally, the costs of the training were evaluated using a micro-costing procedure. RESULTS: Twenty-one out of the 24 trainees managed to attain the predefined mastery level two times consecutively. The trainees completed their training with a median of 2h38min (range: 1h20min-4h30min) using a median of 7 attempts (range: 3-11 attempts) at the simulator test. The cost of training one trainee to the mastery level was estimated to be USD 638. CONCLUSION: Complete trainees can obtain mastery learning levels in general abdominal ultrasound examinations within 3 hours of training in the simulated setting and at an average cost of USD 638 per trainee. Future studies are needed to explore how the cost of simulation-based training is best balanced against the costs of clinical training.

Gap Balancing Throughout the Arc of Motion With Navigated TKA and a Novel Force-Controlled Distractor: A Review of the First 273 Cases.

BACKGROUND: This study evaluated the ability to achieve the targeted soft-tissue balance in terms of medio-lateral (ML) laxity and gap values when using a computer-assisted orthopedic surgery (CAOS) system featuring an intra-articular force-controlled distractor and assessed learning curves associated with the adoption of this technology. METHODS: The first 273 cases using this technology were reported without exclusions comparing 1) final ML laxity and 2) final average gap to their predefined targets. For both parameters, the signed and unsigned differentials were reported. The linear mixed model was used to evaluate laxity curve differences between surgeons. A cumulative sum control chart (CUSUM) was applied to assess surgeon learning curves regarding surgical time. RESULTS: Both the average signed ML laxity and gap differentials were neutral throughout the full arc of motion. Both the average unsigned ML laxity and gap differentials were linear. Signature of ML laxity and gap differential curves tended to be surgeon-specific. The CUSUM analyses of surgical times demonstrated either a short learning curve or the absence of a discernible learning pattern for surgeons. CONCLUSION: Data from all users involved with the pilot release of the balancing device were considered to capture variability in familiarity with the technique and learning curve cases were included. A high ability to achieve targeted gap balance throughout the arc of motion using the proposed method was observed.

Differences in training among prehospital emergency physicians in Germany.

Background: Germany has an interdisciplinary physician-based emergency medical service. Differences in training likely lead to different levels of expertise. Objectives: We assessed the number of manual procedures performed at the completion of training to determine level of experience of prehospital emergency physicians of different primary specialties. Materials and methods: Immediately after passing the board examination each examinee was asked to estimate the number of performed procedures for 26 manual skills. We compared the results with recommendations and data on learning manual skills. Results are presented as mean (standard deviation). Results: Endotracheal intubation via direct laryngoscopy was performed 1032 (739) times by anesthesiologists. Surgeons and internists performed 89 (89) and 77 (65) intubations, respectively. Intubation via video laryngoscopy was performed 79 (81) times by anesthesiologists, 11 (17) times by surgeons and 6 (11) times by internists. Surgeons had little experience in non-invasive ventilation, with 9 (19) performed procedures and had rarely used external pacemaker therapy or electrical cardioversion. In comparison, among all participants non-invasive ventilation was performed 152 (197) times, electrical cardioversion was performed 41 (103) times and an external pacemaker was used 6 (15) times. For other procedures the numbers did not markedly differ between the different specialties. Conclusion: The number of performed procedures markedly differed for some skills between different primary specialties. Recommendations regarding a procedural volume were not always met, suggesting missing expertise for some skills. A defined number of procedures should therefore be a formal requirement to be eligible for board certification in prehospital emergency medicine.

Multi-level longitudinal learning curve regression models integrated with item difficulty metrics for deliberate practice of visual diagnosis: groundwork for adaptive learning.

Visual diagnosis of radiographs, histology and electrocardiograms lends itself to deliberate practice, facilitated by large online banks of cases. Which cases to supply to which learners in which order is still to be worked out, with there being considerable potential for adapting the learning. Advances in statistical modeling, based on an accumulating learning curve, offer methods for more effectively pairing learners with cases of known calibrations. Using demonstration radiograph and electrocardiogram datasets, the advantages of moving from traditional regression to multilevel methods for modeling growth in ability or performance are demonstrated, with a final step of integrating case-level item-response information based on diagnostic grouping. This produces more precise individual-level estimates that can eventually support learner adaptive case selection. The progressive increase in model sophistication is not simply statistical but rather brings the models into alignment with core learning principles including the importance of taking into account individual differences in baseline skill and learning rate as well as the differential interaction with cases of varying diagnosis and difficulty. The developed approach can thus give researchers and educators a better basis on which to anticipate learners' pathways and individually adapt their future learning.

Implementation of transbronchial lung cryobiopsy in a tertiary referral center for interstitial lung diseases: a cohort study on diagnostic yield, complications, and learning curves.

BACKGROUND: Transbronchial lung cryobiopsy (TBLC) has been introduced as an alternative to surgical lung biopsy (SLB) in the diagnostics of interstitial lung diseases (ILD). Despite controversy on safety, TBLC is increasingly implemented in ILD centers with an apparent diagnostic yield comparable to SLB. The aim of this study was to assess TBLC implementation experiences from a tertiary Danish ILD center regarding diagnosis, complications, and learning curves for TBLC performance. METHODS: TBLC was prospectively performed in a cohort of patients with unclassifiable ILD based on a preceding multidisciplinary clinical and radiological revision. TBLC was performed as an outpatient procedure with the patients in general anesthesia using a flexible bronchoscope with 1.9 or 2.4 mm cryoprobes. Learning curves for TBLC performance were calculated using cumulated sum (CUSUM) scores for diagnostic yield, pneumothorax, and bleeding. RESULTS: From February 2017 to March 2020 141 patients (86 (61%) men, median age 69 years [IQR, 60-74 years]) had TBLC performed. A histological and confirmative diagnosis was made in 101 patients (75.2%) and 124 patients (87.9%, i.e. clinical diagnostic yield), respectively, in whom idiopathic interstitial pneumonias constituted the majority (67.3%) of the clinical diagnoses. We observed 2 deaths (1.4%) within 30 days of TBLC, but no procedure-related mortality or severe bleeding. Moderate bleeding occurred in 23 patients (16.3%), pneumothorax in 21 patients (14.9%) with only 14 patients (9.9%) requiring a pleural drain. Based on the CUSUM score analysis, the diagnostic yield obtained was satisfactory throughout the period. CONCLUSION: This study reports experiences of outpatient TBLC implementation in a tertiary referral ILD center from the largest investigated TBLC cohort in Scandinavia The diagnostic yield and prevalence of complications obtained by TBLC from this single center study on unclassifiable ILD support outpatient TBLC as a valuable and safe alternative to SLB to diagnose ILD in well-selected patients. The learning curves for TBLC were acceptable in the hands of experienced bronchoscopists.

Practice Effects of Mobile Tests of Cognition, Dexterity, and Mobility on Patients With Multiple Sclerosis: Data Analysis of a Smartphone-Based Observational Study.

BACKGROUND: Smartphones and their built-in sensors allow for measuring functions in disease-related domains through mobile tests. This could improve disease characterization and monitoring, and could potentially support treatment decisions for multiple sclerosis (MS), a multifaceted chronic neurological disease with highly variable clinical manifestations. Practice effects can complicate the interpretation of both improvement over time by potentially exaggerating treatment effects and stability by masking deterioration. OBJECTIVE: The aim of this study is to identify short-term learning and long-term practice effects in 6 active tests for cognition, dexterity, and mobility in user-scheduled, high-frequency smartphone-based testing. METHODS: We analyzed data from 264 people with self-declared MS with a minimum of 5 weeks of follow-up and at least 5 repetitions per test in the Floodlight Open study, a self-enrollment study accessible by smartphone owners from 16 countries. The collected data are openly available to scientists. Using regression and bounded growth mixed models, we characterized practice effects for the following tests: electronic Symbol Digit Modalities Test (e-SDMT) for cognition; Finger Pinching and Draw a Shape for dexterity; and Two Minute Walk, U-Turn, and Static Balance for mobility. RESULTS: Strong practice effects were found for e-SDMT (n=4824 trials), Finger Pinching (n=19,650), and Draw a Shape (n=19,019) with modeled boundary improvements of 40.8% (39.9%-41.6%), 86.2% (83.6%-88.7%), and 23.1% (20.9%-25.2%) over baseline, respectively. Half of the practice effect was reached after 11 repetitions for e-SDMT, 28 repetitions for Finger Pinching, and 17 repetitions for Draw a Shape; 90% was reached after 35, 94, and 56 repetitions, respectively. Although baseline performance levels were highly variable across participants, no significant differences between the short-term learning effects in low performers (5th and 25th percentile), median performers, and high performers (75th and 95th percentile) were found for e-SDMT up to the fifth trial (ß=1.50-2.00). Only small differences were observed for Finger Pinching (ß=1.25-2.5). For U-Turn (n=15,051) and Static Balance (n=16,797), only short-term learning effects could be observed, which ceased after a maximum of 5 trials. For Two Minute Walk (n=14,393), neither short-term learning nor long-term practice effects were observed. CONCLUSIONS: Smartphone-based tests are promising for monitoring the disease trajectories of MS and other chronic neurological diseases. Our findings suggest that strong long-term practice effects in cognitive and dexterity functions have to be accounted for to identify disease-related changes in these domains, especially in the context of personalized health and in studies without a comparator arm. In contrast, changes in mobility may be more easily interpreted because of the absence of long-term practice effects, even though short-term learning effects might have to be considered.

Learning curves in radiological reporting of whole-body MRI in plasma cell disease: a retrospective study.

BACKGROUND: The plasma cell disease is been studying by the whole-body MRI technology. However, the time requested to learn this radiological technique is unknown. PURPOSE: To esteem, quantitatively and qualitatively, the essential time to learn the whole-body MRI diffusion-weighted imaging with background body signal suppression in patients with plasma cell disease. MATERIALS AND METHODS: Between January 2015 and February 2017, three readers in-training with different levels of experience examined the anonymised and randomised whole-body MRI images of 52 patients with a diagnosis of plasma cell disease and analysed their morphological (T1w, T2w with and without fat suppression) and functional sequences. Reports of an expert radiologist were considered the standard of reference. Images were analysed in two sessions, during which each reader was timed. Readers reported the number of segments with lesions and staged the disease using the Durie-Salmon PLUS staging system. Weighted Cohen's ĸ and Z-test were used to compare the trainees' reports with those of the expert radiologist, and learning curves were drawn up to show changes between the two sessions. RESULTS: Weighted Cohen's ĸ of number of lesioned segments increased from 0.536 ± 0.123 to 0.831 ± 0.129 (Prob > Z under 0.005), thus approaching the goal of ĸ > 0.8. Trainees reached the level of experienced radiologist in terms of time by the 33rd patient. Agreement concerning the Durie-Salmon PLUS increased from 0.536 ± 0.123 to 0.831 ± 0.129 (Prob > Z under 0.005). CONCLUSIONS: The findings of this study demonstrate that whole-body MRI with DWIBS can be learned in about 80 reports and leads to a high level of inter-observer concordance when using the Durie-Salmon PLUS staging system.

Overview of Machine Learning Process Modelling.

Much research has been conducted in the area of machine learning algorithms; however, the question of a general description of an artificial learner's (empirical) performance has mainly remained unanswered. A general, restrictions-free theory on its performance has not been developed yet. In this study, we investigate which function most appropriately describes learning curves produced by several machine learning algorithms, and how well these curves can predict the future performance of an algorithm. Decision trees, neural networks, Naïve Bayes, and Support Vector Machines were applied to 130 datasets from publicly available repositories. Three different functions (power, logarithmic, and exponential) were fit to the measured outputs. Using rigorous statistical methods and two measures for the goodness-of-fit, the power law model proved to be the most appropriate model for describing the learning curve produced by the algorithms in terms of goodness-of-fit and prediction capabilities. The presented study, first of its kind in scale and rigour, provides results (and methods) that can be used to assess the performance of novel or existing artificial learners and forecast their 'capacity to learn' based on the amount of available or desired data.