Deep brain stimulation (DBS) is an effective treatment for Parkinson's disease, but the extent of improvement in motor symptoms varies. A tool which accurately predicts patient outcomes based on data available pre-surgery would be useful for clinical decision-making and patient expectation management. Such data includes clinical and cognitive measures, neuroimaging, kinematics, functional connectivity and genetics. In this systematic review, we assess predictive models of motor outcomes from DBS. We searched the databases Web of Science, PubMed and Scopus for primary research articles that tested predictions from models based on pre-surgical data and focused on motor outcomes from DBS for Parkinson's disease. We identified 19 studies fitting these criteria. The studies with high statistical power and generalisability use only clinical data and have limited accuracy. Studies including other types of data, such as magnetic resonance imaging, may have high accuracy but are under-powered. Predictions are mostly limited to the first year after surgery, subthalamic nucleus-targeted DBS and sum scores of motor performance. Following the results of the systematic search, we discuss candidate input and output variables and validation strategies to produce predictive models that are ready for translation to clinical practice. Ideally, models would predict scores for several motor domains, over a range of time after surgery, with confidence intervals. They should also be generalisable to clinics worldwide. To achieve this goal, models and datasets should be made publicly available to enable wider validation. These recommendations provide a framework to achieve predictive models for DBS outcomes that can be used clinically.

Accurate segmentation of subcortical regions is critical for neurosurgical planning and functional research. Most automated methods rely on template space coregistration, which may compromise patient-specific accuracy, particularly in small structures. We identify a need to evaluate whether native space approaches offer a measurable advantage, which we evaluate in the context of movement disorders. We developed two UNet-based segmentation pipelines of the Subthalamic Nucleus (STN) - a common surgical target in Parkinson's Disease - and the neighbouring Red Nucleus (RN) and Substantia Nigra (SN). We collected 7T and 3T MRI data from five public datasets. The pipelines were evaluated in the native-space against manual labels. We further investigated the effect of the template resolution. Motivated by the hypothesis that models may better learn target boundaries in higher field, we tested the transferability of 7T-trained models to 3T clinical images, and whether synthetic 3T training data - generated via a disentangled representation learning method - could help bridging this domain gap. On held-out 7T data, the native pipeline consistently outperformed the template one. For the STN, native-space Dice reached 0.775 +- 0.055 versus 0.713 +- 0.051 (1 mm template), with HD95 of 0.79 +- 0.24 mm versus 1.17 +- 1.10 mm, respectively. Similar advantages were observed for the RN and SN. Increasing template resolution did not improve accuracy. When applied to 3T images, all models showed a considerable performance drop. Adding synthetic 3T data yielded only modest improvements, though without degrading 7T performance. Native-space segmentation is preferable for applications requiring patient specific anatomical fidelity, such as the surgical planning in PD. Bridging the 7T-to-3T domain gap remains an open challenge, motivating future work on domain adaptation tailored to subcortical structures.

We present a multimodal framework to improve the precision of electrode placement in deep brain stimulation (DBS) by fusing preoperative neuroimaging with intraoperative electrophysiology for accurate electrode co-alignment. The workflow integrates automated subthalamic nucleus (STN) segmentation from preoperative MRI using a two-step convolutional neural network (CNN), classification of microelectrode recordings (MER) with a transformer encoder and spatial co-alignment via a discrete optimisation procedure. Implemented as a 3D Slicer plugin, the pipeline enables real-time visualisation and interactive use during surgery. In validation on retrospective data of 17 trajectories from 12 Parkinson's disease patients, co-alignment reduced the mean lateral localisation error by 0.3 mm relative to an intraoperative reference, indicating improved agreement between electrophysiological and anatomical targets. Automated STN segmentation achieved a Dice similarity of 0.62 ± 0.10, providing a robust starting point for manual refinement. This approach improves the understanding of electrode position within STN during surgery, incorporating preoperative and intraoperative data, offers clinicians a practical, real-time tool to enhance targeting accuracy. By directly integrating imaging and MER evidence, the framework addresses persistent challenges in DBS and represents a step toward more personalised and precise neurosurgical interventions.

Introduction Deep brain stimulation (DBS) of the internal globus pallidus (GPi) is a well-established, effective treatment for dystonia. Substantial variability of therapeutic success has been the one of the drivers of an ongoing debate about proper stimulation site and settings, with several indications of the notional sweet spot pointing to the lower GPi or even subpallidal area. Methods The presented patient-blinded, random-order study with cross-sectional verification against healthy controls enrolled 17 GPi DBS idiopathic, cervical or generalised dystonia patients to compare the effect of the stimulation in the upper and lower GPi area, with the focus on sensorimotor network connectivity and local activity measured using functional magnetic resonance. Results Stimulation brought both these parameters to levels closer to the state detected in healthy controls. This effect was much more pronounced during the stimulation in the lower GPi area or beneath it than in slightly higher positions, with stimulation-related changes detected by both metrics of interest in the sensorimotor cortex, striatum, thalamus and cerebellum. Conclusions All in all, this study not only replicated the results of previous studies on GPi DBS as a modality restoring sensorimotor network connectivity and local activity in dystonia towards the levels in healthy population, but also showed that lower GPi area or even subpallidal structures, be it white matter or even small, but essential nodes in the zona incerta as nucleus basalis of Meynert, are important regions to consider when programming DBS in dystonia patients.

Background:Optimisation of Deep Brain Stimulation (DBS) settings is a key aspect in achieving clinical efficacy in movement disorders, such as the Parkinson's disease. Modern techniques attempt to solve the problem through data-intensive statistical and machine learning approaches, adding significant overhead to the existing clinical workflows. Here, we present a geometry-based optimisation approach for DBS electrode contact and current selection, grounded in routinely collected MRI data, well-established tools (Lead-DBS) and optionally, clinical review records. Methods:The pipeline, packaged in a cross-platform tool, uses lead reconstruction data and simulation of Volume of Tissue Activated (VTA) to estimate the contacts in optimal position relative to the target structure, and suggests optimal stimulation current. The tool then allows further interactive user optimisation of the current settings. Existing electrode contact evaluations can be optionally included in the calculation process for further fine-tuning and adverse effect avoidance. Results:Based on a sample of 174 implanted electrode reconstructions from 87 Parkinson's disease patients, we demonstrate that our algorithm's DBS parameter settings are more effective in covering the target structure (Wilcoxon p < 5e-13, Hedges' g > 0.94) and minimising electric field leakage to neighbouring regions (p < 2e-10, g > 0.46) compared to expert parameter settings. Retrospective analysis of a limited subset (n = 50) predicts comparable improved motor outcomes with expert settings (g = 0.05-0.08, p = 0.09-1), suggesting potential for similar clinical efficacy, pending prospective validation. Conclusion:The proposed automated method for optimisation of the DBS electrode contact and current selection shows promising results and is readily applicable to existing clinical workflows. We demonstrate that the algorithmically selected contacts perform better than manual selections according to electric field calculations, without the iterative optimisation procedure.

Improving the controllability, portability, and inference speed of diffusion language models (DLMs) is a key challenge in natural language generation. While recent research has shown significant success in complex text generation with language models, the memory and computational power are still very demanding and fall short of expectations, which naturally results in low portability and instability for the models. To mitigate these issues, numerous well-established methods were proposed for neural network quantization. To further enhance their portability of independent deployment as well as improve their stability evaluated by language perplexity, we propose a novel approach called the Quantized Embedding Controllable Diffusion Language Model (QE-CDLM). QE-CDLM builds upon the recent successful controllable DLMs by remodeling the task-specific embedding space via quantization. This leads to a gradient-based controller for the generation tasks, and more stable intermediate latent variables are obtained, which naturally brings in an accelerated convergence as well as better controllability. Additionally, the adaption fine-tuning method is employed to reduce tunable weights. Experimental results on five challenging fine-grained control tasks demonstrate that QE-CDLM compares favorably to existing methods in terms of quality and feasibility, achieving better perplexity and lightweight fine-tuning.

Fine-tuning pre-trained language models (LMs) may not always be the most practical approach for downstream tasks. While adaptation fine-tuning methods have shown promising results, a clearer explanation of their mechanisms and further inhibition of the transmission of information is needed. To address this, we propose an Inhibition Adaptation (InA) fine-tuning method that aims to reduce the number of added tunable weights and appropriately reweight knowledge derived from pre-trained LMs. The InA method involves (1) inserting a small trainable vector into each Transformer attention architecture and (2) setting a threshold to directly eliminate irrelevant knowledge. This approach draws inspiration from the shunting inhibition, which allows the inhibition of specific neurons to gate other functional neurons. With the inhibition mechanism, InA achieves competitive or even superior performance compared to other fine-tuning methods on BERT-large, RoBERTa-large, and DeBERTa-large for text classification and question-answering tasks.

To explore the feasibility of avoiding the confident error (or hallucination) of generation models (GMs), we for-malise the system of GMs as a class of stochastic dynamical systems through the lens of control theory. Numerous factors can be attributed to the hallucination of the learning process of GMs, utilising knowledge of control theory allows us to analyze their system functions and system responses. Due to the high complexity of GMs when using various optimization methods, we cannot figure out their solution of Laplace transform, but from a macroscopic perspective, simulating the source response provides a virtual way to address the hallucination of GMs. We also find that the training progress is consistent with the corresponding system response, which offers us a useful way to develop a better optimization component. Finally, the hallucination problem of GMs is fundamentally optimized by using Laplace transform analysis.

BackgroundSerious games, which are gaming applications used for purposes beyond entertainment to educate users on, and address, specific issues, may present a timely approach to promote healthy diabetes management behaviors among children with type 1 diabetes mellitus (T1DM). The lasting benefits associated with these serious games encompass improved patient education; enhanced glycemic control; the reinforcement of bonds within the community of people with diabetes; the facilitation of meaningful dialogues with caregivers, especially within the familial setting; and a significant reduction in the economic burdens associated with subsequent complications. ObjectiveThis paper primarily aims to provide a detailed overview of the iterative design process and the associated evaluation methods used in the development of the educational game. Furthermore, this study aims to enhance motivation for sustained and extended engagement with the game over time. The MyDiabetic game design aims to educate children on various aspects, including the connections among food, insulin, and physical activity. Furthermore, it seeks to impart knowledge related to the operation of a glucometer and an insulin pen, as well as more advanced technologies such as administering glucagon, measuring ketoacidosis, and continuous glucose monitoring. MethodsThe co-design methodology was applied, involving interviews, design workshops, and prototype feedback sessions. A combination of several approaches, such as tailoring, observational learning, social and family support, decision-making practice, and reward systems, was used to support children’s compliance. Moreover, incorporating the literature, guidelines, and current practices into the design ensured that the game was aligned with established health care pathways and included relevant information and best practices for diabetes management. ResultsThe game was tested on 32 children in 3 iterations. Positive responses were received from children who tested the game as well as their parents. The game was also presented to 5 schoolmates of children with T1DM who appreciated a better understanding of the disease and the opportunity to support their friends more efficiently in T1DM compensation. The involvement of children and clinicians in participatory co-design contributed to to the game's high acceptance. With regard to the game’s impact on education, 1 week of testing revealed an enhancement in educational outcomes. ConclusionsThe game is especially suitable for children newly diagnosed with T1DM because it acquaints them in a fun way with new terminology; for example, they can try to measure glycemia levels in an interactive way. The game also caters to children who still need to develop reading skills by including an audio guide. The guide ensures that children of all literacy levels can benefit from the game’s educational content and interactive experiences. The game is available for download on Google Play and the Apple App Store.

Background Deep Brain Stimulation (DBS), applying chronic electrical stimulation of subcortical structures, is a clinical intervention applied in major neurologic disorders. In order to achieve a good clinical effect, accurate electrode placement is necessary. The primary localisation is typically based on presurgical MRI imaging, often followed by intra-operative electrophysiology recording to increase the accuracy and to compensate for brain shift, especially in cases where the surgical target is small, and there is low contrast: e.g., in Parkinson’s disease (PD) and in its common target, the subthalamic nucleus (STN). Methods We propose a novel, fully automatic method for intra-operative surgical navigation. First, the surgical target is segmented in presurgical MRI images using a statistical shape-intensity model. Next, automated alignment with intra-operatively recorded microelectrode recordings is performed using a probabilistic model of STN electrophysiology. We apply the method to a dataset of 120 PD patients with clinical T2 1.5T images, of which 48 also had available microelectrode recordings (MER). Results The proposed segmentation method achieved STN segmentation accuracy around dice = 0.60 compared to manual segmentation. This is comparable to the state-of-the-art on low-resolution clinical MRI data. When combined with electrophysiology-based alignment, we achieved an accuracy of 0.85 for correctly including recording sites of STN-labelled MERs in the final STN volume. Conclusion The proposed method combines image-based segmentation of the subthalamic nucleus with microelectrode recordings to estimate their mutual location during the surgery in a fully automated process. Apart from its potential use in clinical targeting, the method can be used to map electrophysiological properties to specific parts of the basal ganglia structures and their vicinity.

Large language models (LLMs) have demonstrated impressive generalization capabilities on specific tasks with human-written instruction data. However, the limited quantity, diversity, and professional expertise of such instruction data raise concerns about the performance of LLMs in psychotherapy tasks when provided with domain-specific instructions. To address this, we firstly propose Domain-Specific Assistant Instructions based on AlexanderStreet therapy, and secondly we use an adaption fine-tuning method and retrieval augmented generation method to improve pre-trained LLMs. Through quantitative evaluation of linguistic quality using automatic and human evaluation, we observe that pre-trained LLMs on Psychotherapy Assistant Instructions outperform state-of-the-art LLMs response baselines. Our Assistant-Instruction approach offers a half-annotation method to align pre-trained LLMs with instructions, and provide pre-trained LLMs more psychotherapy knowledge.

Deep Brain Stimulation (DBS) is one of the most successful methods to diminish late-stage Parkinson's Disease (PD) symptoms. It is a delicate surgical procedure which requires detailed pre-surgical patient's study. High-field Magnetic Resonance Imaging (MRI) has proven its improved capacity of capturing the Subthalamic Nucleus (STN) - the main target of DBS in PD - in greater detail than lower field images. Here, we present a comparison between the performance of two different Deep Learning (DL) automatic segmentation architectures, one based in the registration to a brain template and the other performing the segmentation in in the MRI acquisition native space. The study was based on publicly available high-field 7 Tesla (T) brain MRI datasets of T1-weighted and T2-weighted sequences. nnUNet was used on the segmentation step of both architectures, while the data pre and post-processing pipelines diverged. The evaluation metrics showed that the performance of the segmentation directly in the native space yielded better results for the STN segmentation, despite not showing any advantage over the template-based method for the to other analysed structures: the Red Nucleus (RN) and the Substantia Nigra (SN).

Major Depressive Disorder (MDD) - can be evaluated by advanced neurocomputing and traditional machine learning techniques. This study aims to develop an automatic system based on a Brain-Computer Interface (BCI) to classify and score depressive patients by specific frequency bands and electrodes. In this study, two Residual Neural Networks (ResNets) based on electroencephalogram (EEG) monitoring are presented for classifying depression (classifier) and for scoring depressive severity (regression). Significant frequency bands and specific brain regions are selected to improve the performance of the ResNets. The algorithm, which is estimated by 10-fold cross-validation, attained an average accuracy rate ranging from 0.371 to 0.571 and achieved average Root-Mean-Square Error (RMSE) from 7.25 to 8.41. After using the beta frequency band and 16 specific EEG channels, we obtained the best-classifying accuracy at 0.871 and the smallest RMSE at 2.80. It was discovered that signals extracted from the beta band are more distinctive in depression classification, and these selected channels tend to perform better on scoring depressive severity. Our study also uncovered the different brain architectural connections by relying on phase coherence analysis. Increased delta deactivation accompanied by strong beta activation is the main feature of depression when the depression symptom is becoming more severe. We can therefore conclude that the model developed here is acceptable for classifying depression and for scoring depressive severity. Our model can offer physicians a model that consists of topological dependency, quantified semantic depressive symptoms and clinical features by using EEG signals. These selected brain regions and significant beta frequency bands can improve the performance of the BCI system for detecting depression and scoring depressive severity.

In this paper, we propose a Fuzzy Window with the Gaussian Processed Label (FW-GPL) method to mitigate the overlap problem in the neighboring ordinal category when scoring images. Many published conventional methods treat this challenge as a traditional regression problem and make a strong assumption that each ordinal category owns an adequate intrinsic rank to outline its distribution. Our FW-GPL method aims to refine the ordinal label pattern by using two novel techniques: (1) assembling fuzzy logic to the fully connected layer of convolution neural networks and (2) transforming the ordinal labels with a Gaussian process. Specifically, it incorporates a heuristic fuzzy logic from the ordinal characteristic and simultaneously plugs in ordinal distribution shapes that penalize the difference between the targeted label and its neighbors to ensure a concentrated regional distribution. Accordingly, the function of these proposed windows is leveraged to minimize the influence of majority classes that mislead the prediction of minority samples. Our model is specifically designed to carefully avoid partially missing continuous facial-age segments. It can perform competitively when using the whole continuous facial-age dataset. Extensive experimental results on three facial-aging datasets and one ambiguous medical dataset demonstrate that our FW-GPL can achieve compelling performance results compared to the State-Of-The-Art (SOTA).

Large language models (LLMs) have demonstrated impressive generalization capabilities throughfine-tuning on specific tasks with human-written instruction data. However, the limited quantity,diversity, and professional expertise of such instruction data raise concerns about the performanceof LLMs in psychotherapy tasks when provided with domain-specific instructions. To addressthis, we propose Domain-Specific Assistant Instructions based on AlexanderStreet therapy andcounseling data and fine-tune pre-trained LLMs on this dataset. Through quantitative evaluation oflinguistic quality using automatic and human evaluation, we observe that pre-trained LLMs finetunedon Psychotherapy Assistant Instructions outperform state-of-the-art LLMs response baselines.Our Assistant-Instruction approach offers a half-annotation method to align pre-trained LLMs withinstructions. We also release our large synthetic dataset, facilitating future studies on professionalinstruction tuning.

eHealth interventions can help people change behavior (eg, quit smoking). Reminders sent via SMS text messaging or email may improve the adherence to web-based programs and increase the probability of successful behavior change; however, it is unclear whether their efficiency is affected by the modality of the communication channel. A 2-armed randomized control trial was conducted to compare the effect of providing reminders via SMS text messaging versus email on the adherence to an eHealth program for smoking cessation and on the probability to initiate a quit attempt. Smokers were recruited via an internet-based advertisement. A total of 591 participants who diverted from intended use of the program (ie, failed to log on to a session) were automatically randomized to the experimental (SMS text messaging reminder, n=304) or the active comparator (email reminder, n=287) group. Unexpectedly, we found that the mode of reminder delivery did not significantly affect either the adherence, namely the number of completed program sessions, with the SMS text messaging reminder group showing a mean of 4.30 (SD 3.24) and the email reminder group showing a mean of 4.36 (SD 3.27) (t586=0.197, P=.84, and Cohen d=0.016), or the outcome, namely the quit smoking attempt rate (34.2% in the SMS text messaging group vs 31.7% in the email group; χ21=0.4, P=.52). Secondary analyses showed that age, gender, and education had significant effects on program adherence and education on the outcome. Moreover, we found a significant interaction effect between the mode of reminder delivery and gender on program adherence, suggesting that the effectiveness of SMS text message reminders might be different for females and males. However, this particular finding should be treated with care as it was based on post hoc subgroup analysis. This study indicates that the modality of user reminders to log on increased neither the program adherence nor the probability of quitting smoking. This suggests that program developers may save costs using emails instead of SMS text messaging reminders. ClinicalTrials.gov NCT03276767; https://clinicaltrials.gov/ct2/show/ NCT03276767

BACKGROUND:This paper describes the development of a mobile app for diabetes mellitus (DM) control and self-management and presents the results of long-term usage of this system in the Czech Republic. DM is a chronic disease affecting large numbers of people worldwide, and this number is continuously increasing. There is massive potential to increase adherence to self-management of DM with the use of smartphones and digital therapeutics interventions.OBJECTIVE:This study aims to describe the process of development of a mobile app, called Mobiab, for DM management and to investigate how individual features are used and how the whole system benefits its long-term users. Using at least 1 year of daily records from users, we analyzed the impact of the app on self-management of DM.METHODS:We have developed a mobile app that serves as an alternative form to the classic paper-based protocol or diary. The development was based on cooperation with both clinicians and people with DM. The app consists of independent individual modules. Therefore, the user has the possibility to use only selected features that they find useful. Mobiab was available free of charge on Google Play Store from mid-2014 until 2019. No targeted recruitment was performed to attract users.RESULTS:More than 500 users from the Czech Republic downloaded and signed up for the mobile app. Approximately 80% of the users used Mobiab for less than 1 week. The rest of the users used it for a longer time and 8 of the users produced data that were suitable for long-term analysis. Additionally, one of the 8 users provided their medical records, which were compared with the gathered data, and the improvements in their glucose levels and overall metabolic stability were consistent with the way in which the mobile app was used.CONCLUSIONS:The results of this study showed that the usability of a DM-centered self-management smartphone mobile app and server-based systems could be satisfactory and promising. Nonetheless, some better ways of motivating people with diabetes toward participation in self-management are needed. Further studies involving a larger number of participants are warranted to assess the effect on long-term diabetes management.

BACKGROUND Self-reported mood is a valuable clinical data source regarding disease state and course in patients with mood disorders. However, validated, quick, and scalable digital self-report measures that can also detect relapse are still not available for clinical care. OBJECTIVE In this study, we aim to validate the newly developed ASERT (Aktibipo Self-rating) questionnaire—a 10-item, mobile app–based, self-report mood questionnaire consisting of 4 depression, 4 mania, and 2 nonspecific symptom items, each with 5 possible answers. The validation data set is a subset of the ongoing observational longitudinal AKTIBIPO400 study for the long-term monitoring of mood and activity (via actigraphy) in patients with bipolar disorder (BD). Patients with confirmed BD are included and monitored with weekly ASERT questionnaires and monthly clinical scales (Montgomery-Åsberg Depression Rating Scale [MADRS] and Young Mania Rating Scale [YMRS]). METHODS The content validity of the ASERT questionnaire was assessed using principal component analysis, and the Cronbach α was used to assess the internal consistency of each factor. The convergent validity of the depressive or manic items of the ASERT questionnaire with the MADRS and YMRS, respectively, was assessed using a linear mixed-effects model and linear correlation analyses. In addition, we investigated the capability of the ASERT questionnaire to distinguish relapse (YMRS≥15 and MADRS≥15) from a nonrelapse (interepisode) state (YMRS<15 and MADRS<15) using a logistic mixed-effects model. RESULTS A total of 99 patients with BD were included in this study (follow-up: mean 754 days, SD 266) and completed an average of 78.1% (SD 18.3%) of the requested ASERT assessments (completion time for the 10 ASERT questions: median 24.0 seconds) across all patients in this study. The ASERT depression items were highly associated with MADRS total scores (P<.001; bootstrap). Similarly, ASERT mania items were highly associated with YMRS total scores (P<.001; bootstrap). Furthermore, the logistic mixed-effects regression model for scale-based relapse detection showed high detection accuracy in a repeated holdout validation for both depression (accuracy=85%; sensitivity=69.9%; specificity=88.4%; area under the receiver operating characteristic curve=0.880) and mania (accuracy=87.5%; sensitivity=64.9%; specificity=89.9%; area under the receiver operating characteristic curve=0.844). CONCLUSIONS The ASERT questionnaire is a quick and acceptable mood monitoring tool that is administered via a smartphone app. The questionnaire has a good capability to detect the worsening of clinical symptoms in a long-term monitoring scenario.