Impact of high-quality, mixed-domain data on the performance of medical language models.

OBJECTIVE: To optimize the training strategy of large language models for medical applications, focusing on creating clinically relevant systems that efficiently integrate into healthcare settings, while ensuring high standards of accuracy and reliability. MATERIALS AND METHODS: We curated a comprehensive collection of high-quality, domain-specific data and used it to train several models, each with different subsets of this data. These models were rigorously evaluated against standard medical benchmarks, such as the USMLE, to measure their performance. Furthermore, for a thorough effectiveness assessment, they were compared with other state-of-the-art medical models of comparable size. RESULTS: The models trained with a mix of high-quality, domain-specific, and general data showed superior performance over those trained on larger, less clinically relevant datasets (P < .001). Our 7-billion-parameter model Med5 scores 60.5% on MedQA, outperforming the previous best of 49.3% from comparable models, and becomes the first of its size to achieve a passing score on the USMLE. Additionally, this model retained its proficiency in general domain tasks, comparable to state-of-the-art general domain models of similar size. DISCUSSION: Our findings underscore the importance of integrating high-quality, domain-specific data in training large language models for medical purposes. The balanced approach between specialized and general data significantly enhances the model's clinical relevance and performance. CONCLUSION: This study sets a new standard in medical language models, proving that a strategically trained, smaller model can outperform larger ones in clinical relevance and general proficiency, highlighting the importance of data quality and expert curation in generative artificial intelligence for healthcare applications.

Nonsurgical Consecutive Exotropia Following Childhood Esotropia: A Multicentered Study.

PURPOSE: Nonsurgical consecutive exotropia (NCX) occurs when an esotropia (ET) spontaneously converts to exotropia (XT) without surgical intervention. Although NCX is considered to occur in early-onset accommodative ET with high hyperopia, consensus on causation is lacking. We report the clinical characteristics of NCX and assess the response to conservative management. DESIGN: Retrospective, multicenter, observational case series. METHODS: Patients aged 6 months and older with an initial diagnosis of ET who converted to XT without surgical intervention. Sensory strabismus was excluded. Age, visual acuity, cycloplegic refraction, glasses prescriptions, deviation, and binocular vision were collected. RESULTS: Forty-nine children were included with a mean age of 3.5 ± 1.6 years and 8.4 ± 3.6 years at the time of ET and NCX, respectively. Mean refractive error was +4.40 ± 2.13 diopters (D) and +4.05 ± 2.74 D at the time of ET and NCX, respectively. Accommodative ET occurred in 60% of cases, and only 35.7% were high hyperopes. All but 1 patient presented with XT at distance. In response to the XT, a mean decrease in hyperopic prescription of 1.55 ± 0.48 D was given (N = 17); only 1 case reverted to ET. Eventually, 43% underwent XT surgery, with similar rates between those who had refractive management and those who did not. CONCLUSIONS: NCX occurs in both accommodative and nonaccommodative ET; high hyperopia is present in only one-third of cases. On average, drift to XT occurs within 5 years. Refractive management has a modest result. No predictive risk factors were identified. Our findings challenge hyperopia-linked theories of causation. Nonrefractive explanations, such as the role of the vergence system, deserve further study.

Spatial Resolution Evaluation Based on Experienced Visual Categories With Sweep Evoked Periodic EEG Activity.

Purpose: Visual function is typically evaluated in clinical settings with visual acuity (VA), a test requiring to behaviorally match or name optotypes such as tumbling E or Snellen letters. The ability to recognize these symbols has little in common with the automatic and rapid visual recognition of socially important stimuli in real life. Here we use sweep visual evoked potentials to assess spatial resolution objectively based on the recognition of human faces and written words. Methods: To this end, we tested unfamiliar face individuation1 and visual word recognition2 in 15 normally sighted adult volunteers with a 68-electrode electroencephalogram system. Results: Unlike previous measures of low-level visual function including VA, the most sensitive electrode was found at an electrode different from Oz in a majority of participants. Thresholds until which faces and words could be recognized were evaluated at the most sensitive electrode defined individually for each participant. Word recognition thresholds corresponded with the VA level expected from normally sighted participants, and even a VA significantly higher than expected from normally sighted individuals for a few participants. Conclusions: Spatial resolution can be evaluated based on high-level stimuli encountered in day-to-day life, such as faces or written words with sweep visual evoked potentials.

Influence of prior and visual information on eye movements in amblyopic children.

This study analyzed the characteristics of pursuit and assessed the influence of prior and visual information on eye velocity and saccades in amblyopic and control children, in comparison to adults. Eye movements of 41 children (21 amblyopes and 20 controls) were compared to eye movements of 55 adults (18 amblyopes and 37 controls). Participants were asked to pursue a target moving at a constant velocity. The target was either a 'standard' target, with a uniform color intensity, or a 'noisy' target, with blurry edges, to mimic the blurriness of an amblyopic eye. Analysis of pursuit patterns showed that the onset was delayed, and the gain was decreased in control children with a noisy target in comparison to amblyopic or control children with a standard target. Furthermore, a significant effect of prior and visual information on pursuit velocity and saccades was found across all participants. Moreover, the modulation of the effect of visual information on the pursuit velocity by group, that is amblyopes or controls with a standard target, and controls with a noisy target, was more limited in children. In other words, the effect of visual information was higher in control adults with a standard target compared to control children with the same target. However, in the case of a blurry target, either in control participants with a noisy target or in amblyopic participants with a standard target, the effect of visual information was larger in children.

Frontotemporal dementia patients exhibit deficits in predictive saccades.

Prediction and time estimation are all but required for motor function in everyday life. In the context of eye movements, for instance, they allow predictive saccades and eye re-acceleration in anticipation of a target re-appearance. While the neural pathways involved are not fully understood, it is known that the frontal lobe plays an important role. As such, neurological disorders that affect it, such as frontotemporal (FTD) dementia, are likely to induce deficits in such movements. In this work, we study the performances of frontotemporal dementia patients in an oculomotor task designed to elicit predictive saccades at different rates, and compare them to young and older adults. Clear deficits in the production of predictive saccades were found in patients, in particular when the time between saccades was short (~500 ms). Furthermore, one asymptomatic C9ORF72 mutation bearer showed patterns of oculomotor behavior similar to FTD patients. He exhibited FTD symptoms within 3 years post-measure, suggesting that an impairment of oculomotor function could be an early clinical sign. Taken together, these results argue in favor of a role of the frontal lobe in predictive movements timing over short timescales, and suggest that predictive saccades in FTD patients warrant further investigation to fully assess their potential as a diagnostic aid.

Vision Screening in Belgian Children: Too Much or Not Enough?

PURPOSE: To describe the vision screening procedures in the three Belgian linguistic communities and present Belgian screening results. METHODS: Analyses were carried out on 636 260 Belgian children between 2009 and 2016. Pre-school children were tested once or twice, while schoolchildren were tested seven or eight times. RESULTS: 57-83% of Belgian pre-school children and close to 100% of school children were tested. Proportions of referrals varied between 9% and 15% for pre-school children and 6-18% for school children. Proportions of failed or doubtful examinations (i.e., incomplete examinations or those with unreliable results) decreased from 1% to 10% in pre-school to 1-4% in school children. Proportions of true positive referrals (36-83%) at the pre-school level varied between linguistic communities. Estimated costs of pre-school screening appeared to be lower in the Flemish community. However, the cost of the ophthalmologist's time per pre-school child with a confirmed diagnostic was probably lower in the French community, as a result of a higher proportion of true referrals. CONCLUSION: Compared to screening programs in 10 developed countries, at the pre-school level, population coverage was lower in the French community, and comparable in the other communities. Proportions of referred children were similar. True positive referrals were comparable in the French community, but below results in other countries in the other communities. At the school level, population coverages were in the higher range of results compared to other countries. Proportions of referrals were below the results of other countries in the Flemish community, and comparable in the rest of the country.

Motor skills in children with strabismus.

PURPOSE: To assess the extent to which strabismus in children was associated with motor difficulties and to examine which parameters of strabismus were most closely associated with motor development. METHODS: The motor skills of children who were suffering from strabismus, were tested binocularly using the Movement Assessment Battery for Children, Second Edition (MABC-2) and compared with the motor performance of monocularly tested healthy controls without any ophthalmologic disease. RESULTS: A total of 40 children with strabismus (mean, 7.25 ± 3.83 years; 19 females) and 18 controls (mean age, 8.33 ± 5.42 years; 6 females) were tested. According to the MABC-2 test, of the 40, 19 had no motor disability, and 21 were at risk of or already presented significant motor disabilities. Results of the MABC-2 were significantly lower for strabismic children without binocularity compared to those with binocularity (P = 0.002). Lack of binocularity was associated with significantly lower performance for static balance (P = 0.003) as well as for catching tasks (P = 0.042). CONCLUSIONS: Lack of binocularity and stereopsis in children is associated with significant motor skills impairment, in particular for static balance and catching tasks. These results should be confirmed with a larger sample, including older patients, to assess the compensation mechanisms that develop with age and the actual effects of strabismus on overall motor performance.

Integration of past and current visual information during eye movements in amblyopia.

Combination of signals based on their reliability is an increasingly popular model for sensorimotor processing. However, how reliability is estimated, or how such estimation is affected by prolonged exposure to noisy inputs, is still unknown. In this study, we compare patients with unilateral functional amblyopia with control subjects tracking either a reliable target, or a blurry, unreliable target, in a task of repeated, sustained smooth pursuit. We provide evidence for a lower weight of visual information during smooth pursuit in amblyopic and control subjects tracking a blurry target, with no significant difference of prior information weight. In contrast, we found no evidence of lower visual information weight in the catch-up saccades of amblyopic subjects. We conclude that oculomotor performance in unilateral amblyopia mostly lays within the continuum between our control groups, without significant differences in the relative weights of prior and visual information. However, smooth pursuit exhibits additional deficits that might result from abnormal visual development.

Rapid Objective Assessment of Contrast Sensitivity and Visual Acuity With Sweep Visual Evoked Potentials and an Extended Electrode Array.

Purpose: Sweep visual evoked potentials (sVEPs) provide an implicit, objective, and sensitive evaluation of low-level visual functions such as visual acuity and contrast sensitivity. For practical and traditional reasons, sVEPs in ophthalmologic examinations have usually been recorded over a single or a limited number of electrodes over the medial occipital region. Here we examined whether a higher density of recording electrodes improves the estimation of individual low-level visual thresholds with sVEPS, and to which extent such testing could be streamlined for clinical application. Methods: To this end, we tested contrast sensitivity and visual acuity in 26 healthy adult volunteers with a 68-electrode electroencephalogram (EEG) system. Results: While the most sensitive electrophysiologic response was found at the traditional medial occipital electrode Oz in a small majority of individuals, it was found at neighboring electrodes for the remaining participants. At the group level, lower spatial frequencies were also associated with right lateralized responses. More generally, visual function was evaluated more sensitively based on EEG recorded at the most sensitive electrode defined individually for each participant. Our data suggest that recording over seven posterior electrodes while limiting the testing session to less than 15 minutes ensures a sensitive and consistent estimation of acuity and contrast sensitivity threshold estimates in every individual. Conclusions: The present study shows that sampling from a larger number of posterior scalp electrodes is relevant to optimize visual function assessment and could be achieved efficiently in the time-constrained clinical setting.

Development of internal models and predictive abilities for visual tracking during childhood.

The prediction of the consequences of our own actions through internal models is an essential component of motor control. Previous studies showed improvement of anticipatory behaviors with age for grasping, drawing, and postural control. Since these actions require visual and proprioceptive feedback, these improvements might reflect both the development of internal models and the feedback control. In contrast, visual tracking of a temporarily invisible target gives specific markers of prediction and internal models for eye movements. Therefore, we recorded eye movements in 50 children (aged 5-19 yr) and in 10 adults, who were asked to pursue a visual target that is temporarily blanked. Results show that the youngest children (5-7 yr) have a general oculomotor behavior in this task, qualitatively similar to the one observed in adults. However, the overall performance of older subjects in terms of accuracy at target reappearance and variability in their behavior was much better than the youngest children. This late maturation of predictive mechanisms with age was reflected into the development of the accuracy of the internal models governing the synergy between the saccadic and pursuit systems with age. Altogether, we hypothesize that the maturation of the interaction between smooth pursuit and saccades that relies on internal models of the eye and target displacement is related to the continuous maturation of the cerebellum.

Spontaneous improvement in oculomotor function of children with cerebral palsy.

Eye movements are essential to get a clear vision of moving objects. In the present study, we assessed quantitatively the oculomotor deficits of children with cerebral palsy (CP). We recorded eye movements of 51 children with cerebral palsy (aged 5-16 years) with relatively mild motor impairment and compared their performance with age-matched control and premature children. Overall eye movements of children with CP are unexpectedly close to those of controls even though some oculomotor parameters are biased by the side of hemiplegia. Importantly, the difference in performance between children with CP and controls decreases with age, demonstrating that the oculomotor function of children with CP develops as fast as or even faster than controls for some visual tracking parameters. That is, oculomotor function spontaneously improves over the course of childhood. This evolution highlights the ability of lesioned brain of children with CP to compensate for impaired motor function beyond what would be achieved by normal development on its own.

The saccadic system does not compensate for the immaturity of the smooth pursuit system during visual tracking in children.

Motor skills improve with age from childhood into adulthood, and this improvement is reflected in the performance of smooth pursuit eye movements. In contrast, the saccadic system becomes mature earlier than the smooth pursuit system. Therefore, the present study investigates whether the early mature saccadic system compensates for the lower pursuit performance during childhood. To answer this question, horizontal eye movements were recorded in 58 children (ages 5-16 yr) and 16 adults (ages 23-36 yr) in a task that required the combination of smooth pursuit and saccadic eye movements. Smooth pursuit performance improved with age. However, children had larger average position error during target tracking compared with adults, but they did not execute more saccades to compensate for their low pursuit performance despite the early maturity of their saccadic system. This absence of error correction suggests that children have a lower sensitivity to visual errors compared with adults. This reduced sensitivity might stem from poor internal models and longer processing time in young children.

Dramatic impairment of prediction due to frontal lobe degeneration.

Prediction is essential for motor function in everyday life. For instance, predictive mechanisms improve the perception of a moving target by increasing eye speed anticipatively, thus reducing motion blur on the retina. Subregions of the frontal lobes play a key role in eye movements in general and in smooth pursuit in particular, but their precise function is not firmly established. Here, the role of frontal lobes in the timing of predictive action is demonstrated by studying predictive smooth pursuit during transient blanking of a moving target in mild frontotemporal lobar degeneration (FTLD) and Alzheimer's disease (AD) patients. While control subjects and AD patients predictively reaccelerated their eyes before the predicted time of target reappearance, FTLD patients did not. The difference was so dramatic (classification accuracy >90%) that it could even lead to the definition of a new biomarker. In contrast, anticipatory eye movements triggered by the disappearance of the fixation point were still present before target motion onset in FTLD patients and visually guided pursuit was normal in both patient groups compared with controls. Therefore, FTLD patients were only impaired when the predicted timing of an external event was required to elicit an action. These results argue in favor of a role of the frontal lobes in predictive movement timing.

Review of the major findings about Duane retraction syndrome (DRS) leading to an updated form of classification.

In view of all the reported evidence by electromyography in the 1970s, by histology in the 1980s, and by cerebral imagery since the 2000s, Duane retraction syndrome (DRS) has been described as the consequence of a congenital anomaly of the 6th cranial nerve nuclei with aberrant innervations by supply from the 3rd cranial nerve. Both genetic and environmental factors are likely to play a role when the cranial nerves and ocular muscles are developing between the 4th and the 8th week of gestation. New data from eye movement recordings contributed to better understanding the binocular control of saccades. Modeling of saccades in DRS seems promising for the quantification of the innervational deficit and the mechanical properties of the eye plant. The usual clinical classification of DRS needs to be updated in order to match more accurately the underlying dysinnervation of the extra ocular muscles and to illustrate the continuum that exists between the various forms. This review aims to summarize the major findings about DRS and to guide the clinician in the surgical management of this particular form of strabismus.

Binocular coordination of saccades in Duane Retraction Syndrome.

Disconjugate oculomotor adaptation is driven by the need to maintain binocular vision. Since binocular vision in Duane Retraction Syndrome (DRS) patients is normal in half of their horizontal field of gaze (i.e., sound-side of gaze), we wondered whether oculomotor adaptive capabilities are efficient despite such a severe impairment of eye motility towards the other half of the horizontal field of gaze (i.e., affected-side gaze). We compared properties of horizontal saccades of patients with congenital unilateral Duane Retraction Syndrome type I in binocular viewing and monocular viewing conditions by simultaneously recording both eyes with the search coil technique. Our results show a mismatch between the pulse and the step signal of the innervation for saccades. When tested in the affected eye viewing condition (sound eye covered), the eyes showed not only similarly-directed increases of the saccadic gain (pulse signal) in the two eyes but also disjunctive post-saccadic drifts (step signal). This behavior suggests that visuomotor errors presented only to the affected eye were transferred to the sound eye, producing conjugate changes of the saccadic command. The post-saccadic command remained unchanged, however, and controlled the final position of each eye separately. This suggests that monocular adaptation is possible only for the step of innervation (i.e., controlling the final eye position) but not for the pulse of innervation (i.e., controlling the saccadic gain), even though the peculiarity of unilateral DRS type I offers a clear advantage for separate pathways of control for the two eyes.

Properties of saccades in Duane retraction syndrome.

PURPOSE: To improve understanding of the binocular control of saccades by making high-resolution eye movement recordings in patients with unilateral Duane retraction syndrome (DRS) type I (marked by congenital absence of the sixth cranial nerve). METHODS: Binocular eye movements were recorded in four patients in binocular viewing conditions during a saccade task. RESULTS: Affected-side gaze showed normal saccades of the sound eye and undershooting saccades of the affected eye-evidence of intact interneurons, but deficient motoneurons, in the sixth-nerve nucleus on the affected side. Postsaccadic smooth onward drift followed undershooting saccades of the affected eye. Sound-side gaze, in the centripetal direction, showed relatively accurate saccades despite large offsets in initial position between the two eyes. In the centrifugal direction, there was a consistent undershoot of the affected eye. The sound eye showed unexpected overshoot. Postsaccadic drifts in opposite directions in the two eyes brought both eyes onto the target in an optimal time course. CONCLUSIONS: Studying the characteristics of the saccadic system in this oculomotor disorder gives new insight into binocular control of saccades and adaptation. In these patients, the oculomotor system was not capable of adapting the pulse command to the two eyes separately, even though DRS offers clear advantages of independent control. In contrast, independent adaptation of the saccadic step command was sent to the two eyes. This supports the idea that the rapid part of the saccadic command is common to both eyes.

What triggers catch-up saccades during visual tracking?

When tracking moving visual stimuli, primates orient their visual axis by combining two kinds of eye movements, smooth pursuit and saccades, that have very different dynamics. Yet, the mechanisms that govern the decision to switch from one type of eye movement to the other are still poorly understood, even though they could bring a significant contribution to the understanding of how the CNS combines different kinds of control strategies to achieve a common motor and sensory goal. In this study, we investigated the oculomotor responses to a large range of different combinations of position error and velocity error during visual tracking of moving stimuli in humans. We found that the oculomotor system uses a prediction of the time at which the eye trajectory will cross the target, defined as the "eye crossing time" (T(XE)). The eye crossing time, which depends on both position error and velocity error, is the criterion used to switch between smooth and saccadic pursuit, i.e., to trigger catch-up saccades. On average, for T(XE) between 40 and 180 ms, no saccade is triggered and target tracking remains purely smooth. Conversely, when T(XE) becomes smaller than 40 ms or larger than 180 ms, a saccade is triggered after a short latency (around 125 ms).