Background Deep neural networks learn from former experiences on a large scale and can be used to predict future disease activity as potential clinical decision support. AdaptiveNet is a novel adaptive recurrent neural network optimized to deal with heterogeneous and missing clinical data. Objective We investigate AdaptiveNet for the prediction of individual disease activity in patients from a rheumatoid arthritis (RA) registry. Methods Demographic and disease characteristics from over 9500 patients and 65.000 visits from the Swiss Quality Management (SCQM) database were used to train and evaluate the network. Patient characteristics, clinical and patient reported outcomes, laboratory values and medication were used as input features. DAS28-BSR served as a target to predict active RA and future numeric individual disease activity by classification and regression. Results AdaptiveNet predicted active disease defined as DAS28-BSR >2.6 at the next visit with an overall accuracy of 75.6% (SD +- 0.7%) and a sensitivity and specificity of 84.2% (SD +- 1.6%) and 61.5% (SD +- 3.6%), respectively. Prediction performance was significantly higher in patients with a disease duration >3 years and positive rheumatoid factor. Regression allowed forecasting individual DAS28-BSR values with a mean squared error (MSE) of 0.9 (SD +- 0.05). This corresponds to a 8% deviation between estimated and real DAS28-BSR values. Compared to linear regression, random forest and support vector machines, AdaptiveNet showed an increased performance of over 7% in MSE. Medication played a minor role in the prediction of RA disease activity. Conclusion AdaptiveNet has a superior capacity to predict numeric RA disease activity compared to classical machine learning architectures. All investigated models had limitations in low specificity.

Digital biomarkers such as wearables are of increasing interest in monitoring rheumatic diseases, but they usually lack disease specificity. In this study, we apply convolutional neural networks (CNN) to real-world hand photographs in order to automatically detect, extract, and analyse dorsal finger fold lines as a correlate of proximal interphalangeal (PIP) joint swelling in patients with rheumatoid arthritis (RA). Hand photographs of RA patients were taken by a smartphone camera in a standardized manner. Overall, 190 PIP joints were categorized as either swollen or not swollen based on clinical judgement and ultrasound. Images were automatically preprocessed by cropping PIP joints and extracting dorsal finger folds. Subsequently, metrical analysis of dorsal finger folds was performed, and a CNN was trained to classify the dorsal finger lines into swollen versus non-swollen joints. Representative horizontal finger folds were also quantified in a subset of patients before and after resolution of PIP swelling and in patients with disease flares. In swollen joints, the number of automatically extracted deep skinfold imprints was significantly reduced compared to non-swollen joints (1.3, SD 0.8 vs. 3.3, SD 0.49, <i>p</i> &#x3c; 0.01). The joint diameter/deep skinfold length ratio was significantly higher in swollen (4.1, SD 1.4) versus non-swollen joints (2.1, SD 0.6, <i>p</i> &#x3c; 0.01). The CNN model successfully differentiated swollen from non-swollen joints based on finger fold patterns with a validation accuracy of 0.84, a sensitivity of 88%, and a specificity of 75%. A heatmap of the original images obtained by an extraction algorithm confirmed finger folds as the region of interest for correct classification. After significant response to disease-modifying antirheumatic drug ± corticosteroid therapy, longitudinal metrical analysis of eight representative deep finger folds showed a decrease in the mean diameter/finger fold length (finger fold index, FFI) from 3.03 (SD 0.68) to 2.08 (SD 0.57). Conversely, the FFI increased in patients with disease flares. In conclusion, automated preprocessing and the application of CNN algorithms in combination with longitudinal metrical analysis of dorsal finger fold patterns extracted from real-world hand photos might serve as a digital biomarker in RA.

Current neural decoding methods typically aim at explaining behavior based on neural activity via supervised learning. However, since generally there is a strong connection between learning of subjects and their expectations on long-term rewards, we propose NeuRL, an inverse reinforcement learning approach that (1) extracts an intrinsic reward function from collected trajectories of a subject in closed form, (2) maps neural signals to this intrinsic reward to account for long-term dependencies in the behavior and (3) predicts the simulated behavior for unseen neural signals by extracting Q-values and the corresponding Boltzmann policy based on the intrinsic reward values for these unseen neural signals. We show that NeuRL leads to better generalization and improved decoding performance compared to supervised approaches. We study the behavior of rats in a responsepreparation task and evaluate the performance of NeuRL within simulated inhibition and per-trial behavior prediction. By assigning clear functional roles to defined neuronal populations our approach offers a new interpretation tool for complex neuronal data with testable predictions. In per-trial behavior prediction, our approach furthermore improves accuracy by up to 15% compared to traditional methods.

Cycling of biologic or targeted synthetic disease modifying antirheumatic drugs (b/tsDMARDs) in rheumatoid arthritis (RA) patients due to non-response is a problem preventing and delaying disease control. We aimed to assess and validate treatment response of b/tsDMARDs among clusters of RA patients identified by deep learning. We clustered RA patients clusters at first-time b/tsDMARD (cohort entry) in the Swiss Clinical Quality Management in Rheumatic Diseases registry (SCQM) [1999–2018]. We performed comparative effectiveness analyses of b/tsDMARDs (ref. adalimumab) using Cox proportional hazard regression. Within 15 months, we assessed b/tsDMARD stop due to non-response, and separately a ≥20% reduction in DAS28-esr as a response proxy. We validated results through stratified analyses according to most distinctive patient characteristics of clusters. Clusters comprised between 362 and 1481 patients (3516 unique patients). Stratified (validation) analyses confirmed comparative effectiveness results among clusters: Patients with ≥2 conventional synthetic DMARDs and prednisone at b/tsDMARD initiation, male patients, as well as patients with a lower disease burden responded better to tocilizumab than to adalimumab (hazard ratio [HR] 5.46, 95% confidence interval [CI] [1.76–16.94], and HR 8.44 [3.43–20.74], and HR 3.64 [2.04–6.49], respectively). Furthermore, seronegative women without use of prednisone at b/tsDMARD initiation as well as seropositive women with a higher disease burden and longer disease duration had a higher risk of non-response with golimumab (HR 2.36 [1.03–5.40] and HR 5.27 [2.10–13.21], respectively) than with adalimumab. Our results suggest that RA patient clusters identified by deep learning may have different responses to first-line b/tsDMARD. Thus, it may suggest optimal first-line b/tsDMARD for certain RA patients, which is a step forward towards personalizing treatment. However, further research in other cohorts is needed to verify our results.

Deciphering the neural code underlying goal-directed behavior is a long-term mission in neuroscience. Neurons exhibiting preparation and movement-related activity are intermingled in the premotor and motor cortices, thus concealing the neural code of planned movements. We employed a combination of electrophysiology, pathway-specific optogenetics, phototagging, and inverse reinforcement learning (RL) to elucidate the role of defined neuronal subpopulations in the rat rostral and caudal forelimb areas (RFA and CFA), which correspond to the premotor and motor cortical areas. The inverse RL enabled the functional dissection of spatially intermingled neuronal subpopulations, complementing our pathway-specific optogenetic manipulations and unveiling differential functions of the preparation and movement subpopulations projecting from RFA to CFA. Our results show that the projecting preparation subpopulation suppresses movements, whereas the projecting movement subpopulation promotes actions. We found the influence of RFA on CFA to be adaptable, with the projection either inhibiting or exciting neurons in the superficial and deep CFA layers, depending on context and task phase. These complex interactions between RFA and CFA likely involve the differential recruitment of inhibitory interneurons in the CFA, which is supported by our electron microscopy analysis of the connectivity between these regions. We provide here unprecedented mechanistic insights into how the premotor and primary motor cortices are functionally and structurally interlinked with the potential to advance neuroprosthetics.