The serotonin hypothesis of depression is still influential. We aimed to synthesise and evaluate evidence on whether depression is associated with lowered serotonin concentration or activity in a systematic umbrella review of the principal relevant areas of research. PubMed, EMBASE and PsycINFO were searched using terms appropriate to each area of research, from their inception until December 2020. Systematic reviews, meta-analyses and large data-set analyses in the following areas were identified: serotonin and serotonin metabolite, 5-HIAA, concentrations in body fluids; serotonin 5-HT 1A receptor binding; serotonin transporter (SERT) levels measured by imaging or at post-mortem; tryptophan depletion studies; SERT gene associations and SERT gene-environment interactions. Studies of depression associated with physical conditions and specific subtypes of depression (e.g. bipolar depression) were excluded. Two independent reviewers extracted the data and assessed the quality of included studies using the AMSTAR-2, an adapted AMSTAR-2, or the STREGA for a large genetic study. The certainty of study results was assessed using a modified version of the GRADE. We did not synthesise results of individual meta-analyses because they included overlapping studies. The review was registered with PROSPERO (CRD42020207203). 17 studies were included: 12 systematic reviews and meta-analyses, 1 collaborative meta-analysis, 1 meta-analysis of large cohort studies, 1 systematic review and narrative synthesis, 1 genetic association study and 1 umbrella review. Quality of reviews was variable with some genetic studies of high quality. Two meta-analyses of overlapping studies examining the serotonin metabolite, 5-HIAA, showed no association with depression (largest n = 1002). One meta-analysis of cohort studies of plasma serotonin showed no relationship with depression, and evidence that lowered serotonin concentration was associated with antidepressant use ( n = 1869). Two meta-analyses of overlapping studies examining the 5-HT 1A receptor (largest n = 561), and three meta-analyses of overlapping studies examining SERT binding (largest n = 1845) showed weak and inconsistent evidence of reduced binding in some areas, which would be consistent with increased synaptic availability of serotonin in people with depression, if this was the original, causal abnormaly. However, effects of prior antidepressant use were not reliably excluded. One meta-analysis of tryptophan depletion studies found no effect in most healthy volunteers ( n = 566), but weak evidence of an effect in those with a family history of depression ( n = 75). Another systematic review ( n = 342) and a sample of ten subsequent studies ( n = 407) found no effect in volunteers. No systematic review of tryptophan depletion studies has been performed since 2007. The two largest and highest quality studies of the SERT gene, one genetic association study ( n = 115,257) and one collaborative meta-analysis ( n = 43,165), revealed no evidence of an association with depression, or of an interaction between genotype, stress and depression. The main areas of serotonin research provide no consistent evidence of there being an association between serotonin and depression, and no support for the hypothesis that depression is caused by lowered serotonin activity or concentrations. Some evidence was consistent with the possibility that long-term antidepressant use reduces serotonin concentration.

Current and next-generation particle tracking detectors will incorporate precision timing capabilities with resolutions approaching tens of picoseconds. Using Technology Computer-Aided Design (TCAD) simulations of Low-Gain Avalanche Diode (LGAD) detectors, we demonstrate that oblique particle incidence induces systematic timing variations of hundreds of picoseconds across multiple pixels. We derive an analytical linear model relating inter-pixel timing differences to incident track angles, enabling single-layer angular reconstruction with few-degree precision. Stochastic energy loss fluctuations (Landau fluctuations) impose fundamental limits on both angular resolution and reconstruction efficiency. Comparison with neural network approaches demonstrates that the linear model achieves near-optimal angular resolution, indicating that the physics of charge collection geometry, rather than algorithmic sophistication, dominates the achievable performance.

We are entering a new era of composite model architectures that integrate diverse components such as vision encoders, language backbones, diffusion and flow heads, audio codecs, action generators, and world-model predictors. Such architectures underpin a broad class of multimodal models, including unified multimodal models, omni models, speech-language models, vision-language-action policies, and world models. However, existing model serving frameworks were built on narrow assumptions about model structure, making them ill-suited to accommodate this new architectural diversity. Here we present M*, a universal serving system for efficient serving of composite AI models. M* represents models as dataflow graphs, processing requests spanning diverse modalities and tasks as traversals over these graphs. The core insight is a modular abstraction that supports arbitrary composition of model components, flexible placement onto a physical cluster, and model-agnostic optimizations within a distributed runtime. We call this abstraction the Walk Graph and show how it can concisely capture composite models from a broad range of families. We instantiate M* on representative models and find that it achieves, on average, 20

Sparsity has recently attracted increased attention in the machine learning (ML) community due to its potential to improve performance and energy efficiency by eliminating ineffectual computations. As ML models evolve rapidly, reconfigurable architectures, such as coarse-grained reconfigurable arrays (CGRAs), are being explored to adapt to and accelerate emerging models. Previous CGRA designs have supported unstructured sparsity and reported promising speedups and energy savings for compute-intensive kernels. However, these approaches still face performance bottlenecks when accelerating entire sparse ML networks. In this letter, we identify the primary sources of inefficiency in prior CGRA-based approaches and present Opal, a CGRA SoC with three key contributions: 1) flexible dataflow architecture supporting Gustavson’s dataflow for sparse matrix multiplication, 2) high-throughput sparse hardware primitives, and 3) enhanced processing elements to support mapping all ML operations on the CGRA. As a result, Opal achieves a 66% to 79% reduction in runtime and energy consumption across our evaluated sparse graph neural network benchmarks compared to prior CGRA solutions which only target kernel acceleration.

Onyx is a system-on-chip (SoC) with a coarse-grained reconfigurable array (CGRA) for accelerating sparse and dense tensor algebra and dense image processing and machine learning (ML) applications. To support multiple inputs, multiple dimensions, and fusion in sparse applications, Onyx utilizes composable memory primitives that operate on compressed storage and streams and compute primitives that eliminate unnecessary calculations. Onyx also improves performance on dense applications with application-specialized processing elements (PEs), area-optimized memory tiles, and hybrid clock gating in the global buffer (GLB). Onyx achieves a peak energy efficiency of 756 INT16 GOPS/W, up to 565x better energy-delay product (EDP) for sparse kernels versus CPUs with sparse libraries, and up to 76% and 85% lower EDP for image processing and ML, respectively, versus the state-of-the-art CGRA.

Abstract Background Mirtazapine is used to treat depression worldwide, and the effects of mirtazapine on depression rating scales are well-known. Our primary objective was to assess the risks of adverse events with mirtazapine for major depressive disorder. Methods We searched relevant sources from inception to 7 March 2024 for randomised clinical trials comparing mirtazapine versus placebo in adults with major depressive disorder. The primary outcomes were suicides or suicide attempts, serious adverse events, and non-serious adverse events. Data were synthesised using meta-analysis and Trial Sequential Analysis. Results We included 17 trials randomising 2,131 participants to mirtazapine versus placebo. All results were at high risk of bias, and the certainty of the evidence was very low. The included trials assessed outcomes at a maximum of 12 weeks after randomisation. Meta-analysis and Trial Sequential Analysis showed insufficient information to determine the effects of mirtazapine on the risks of suicides or suicide attempts and serious adverse events. Meta-analyses showed that mirtazapine increased the risks of somnolence, weight gain, dry mouth, dizziness, and increased appetite but decreased the risk of headaches. Conclusions There is a lack of evidence on the effects of mirtazapine on suicides and serious adverse events. Mirtazapine increases the risks of somnolence, weight gain, dry mouth, dizziness, and increased appetite. Mirtazapine might decrease the risk of headaches. The long-term effects of mirtazapine are unknown. Prospero id CRD42022315395.

Recent research has focused on leveraging sparsity in hardware accelerators to improve the efficiency of applications spanning scientific computing to machine learning. Most such prior accelerators are fixed-function, which is insufficient for two reasons. First, applications typically include both dense and sparse components, and second, the algorithms that comprise these applications are constantly evolving. To address these challenges, we designed a programmable accelerator called Onyx for both sparse tensor algebra and dense workloads. Onyx extends a coarse-grained reconfigurable array (CGRA) optimized for dense applications with composable hardware primitives to support arbitrary sparse tensor algebra kernels. In this article, we show that we can further optimize Onyx by adding a small set of hardware features for parallelization that significantly increase both temporal and spatial utilization of the CGRA, reducing runtime by up to 6.2x.

Objectives To assess the beneficial and harmful effects of duloxetine versus ‘active placebo’, placebo or no intervention for adults with major depressive disorder.Design Systematic review with meta-analysis and trial sequential analysis of randomised trials.Data sources Cochrane Central Register of Controlled Trials, MEDLINE, Embase, PsycINFO and other relevant databases up until January 2023. We requested clinical study reports from 36 competent authorities.Eligibility criteria for selecting studies All randomised clinical trials comparing duloxetine versus placebo, ‘active placebo’ or no intervention, irrespective of publication type, publication status, publication year and language for treatment of major depressive disorder in adults.Data extraction and synthesis Five authors in pairs extracted data using a standardised data extraction sheet. A third review author was consulted for disagreements. Intervention effects were assessed by both random-effects and fixed-effect model meta-analyses, risk of bias assessments were performed by two independent review authors using Cochrane’s risk of bias tool V.2 and the certainty of evidence was assessed using Grading of Recommendations Assessment, Development and Evaluation.Results We included 28 trials randomising a total of 7872 participants. All results were at high risk of bias. The trials’ assessment time points were between 6 and 16 weeks after randomisation. Meta-analyses showed evidence of a beneficial effect of duloxetine on depressive symptoms (mean difference −1.81, Hamilton Depression Rating Scale (HDRS-17) points; 95% CI −2.34 to −1.28; heterogeneity I2=0.0%; 12 trials) and quality of life (mean difference −3.79 points, 95% CI −5.11 to −2.46; I2=0.0%; three trials), but the effect sizes were below our predefined minimal clinically important differences. Trial sequential analysis showed that we did not have enough information to assess the effects of duloxetine on serious adverse events (SAEs) (OR 0.67, 95% CI 0.44 to 1.02; I2=0.0%; 19 trials) or suicide or suicide attempts (OR 1.08, 95% CI 0.37 to 3.16; six trials). Duloxetine increased the risk of non-SAEs (risk ratio 1.27, 95% CI 1.22 to 1.32; I2=73.0%; 24 trials). The adverse events with the lowest number needed to harm (NNH) were nausea (NNH 6), dry mouth (NNH 13), somnolence (NNH 17), withdrawal syndrome (NNH 19), sweating (NNH 20), dizziness (NNH 21) and constipation (NNH 21).Conclusions Duloxetine appears to reduce depressive symptom scores and improve quality of life scores in the short term, but the effect sizes are minimal and of questionable patient importance. The short- and long-term effects of duloxetine on risks of SAEs and suicidality are uncertain. Duloxetine increases the risks of several short-term adverse events. Systematic assessments of benefits and harms over longer periods are required.Trial registration number PROSPERO 2016 CRD42016053931.

BackgroundWe sought to understand more about the nature and possible consequences of antidepressant withdrawal.MethodsWe surveyed members of 20 peer-led, online groups, assessing 31 commonly reported antidepressant withdrawal symptoms.ResultsThere were 1148 respondents, who were mostly white, female and educated. For 40% of respondents, withdrawal symptoms had lasted more than 2 years and 80% were moderately or severely impacted by them. One in four were unable to stop their antidepressant. Reported consequences of withdrawal included impaired work function (56%), losing jobs (20%), taking sick leave (27%), and relationship breakdown (25%). Both emotional and physical symptoms newly occurred or increased in severity following antidepressant withdrawal compared to before starting antidepressants. There was no difference in the nature of symptoms, severity or duration between people with physical or mental health diagnoses. We have proposed a potential Discriminatory Antidepressant Withdrawal Symptoms Scale (DAWSS), comprising the 15 symptoms most specific to withdrawal (including electric shock sensations, dizziness, akathisia or restlessness, vertigo, and vomiting), which requires further validation.LimitationsThe sample was derived from peer support groups and is not representative of everyone who undergoes antidepressant withdrawal. The cross-sectional design precludes establishing causal relationships between variables.ConclusionsOur findings suggest there is a distinctive antidepressant withdrawal syndrome characterised by a range of emotional and physical symptoms, which can be severe, prolonged and have profound impact. The DAWSS may be helpful in distinguishing withdrawal from underlying conditions. Health services need to provide evidence-based clinical advice and support to people on long-term antidepressants.

Question Tricyclic antidepressants are used to treat depression worldwide, but the adverse effects have not been systematically assessed. Our objective was to assess the beneficial and harmful effects of all tricyclic antidepressants for adults with major depressive disorder. Study selection and analysis We conducted a systematic review with meta-analysis and trial sequential analysis. We searched CENTRAL, MEDLINE, Embase, LILACS and other sources from inception to January 2023 for randomised clinical trials comparing tricyclic antidepressants versus placebo or ‘active placebo’ for adults with major depressive disorder. The primary outcomes were depressive symptoms measured on the 17-item Hamilton Depression Rating Scale (HDRS-17), serious adverse events and quality of life. The minimal important difference was defined as three points on the HDRS-17. Findings We included 103 trials randomising 10 590 participants. All results were at high risk of bias, and the certainty of the evidence was very low or low. All trials only assessed outcomes at the end of the treatment period at a maximum of 12 weeks after randomisation. Meta-analysis and trial sequential analysis showed evidence of a beneficial effect of tricyclic antidepressants compared with placebo (mean difference −3.77 HDRS-17 points; 95% CI −5.91 to −1.63; 17 trials). Meta-analysis showed evidence of a harmful effect of tricyclic antidepressants compared with placebo on serious adverse events (OR 2.78; 95% CI 2.18 to 3.55; 35 trials), but the required information size was not reached. Only 2 out of 103 trials reported on quality of life and t-tests showed no evidence of a difference. Conclusions The long-term effects of tricyclic antidepressants and the effects on quality of life are unknown. Short-term results suggest that tricyclic antidepressants may reduce depressive symptoms while also increasing the risks of serious adverse events, but these results were based on low and very low certainty evidence. PROSPERO registration number CRD42021226161.

Objective: To examine the evidence and practice of antipsychotic dose reduction from the lens of biomedical ethics (specifically principlism) to support evidence-based practice and patient choice and self-determination. Methods: An overview of the evidence from randomized controlled trials of antipsychotic dose reduction versus maintenance is presented. This is followed by a theoretical examination of the four key biomedical ethical principles of autonomy, nonmaleficence, beneficence, and justice and how they apply in the case of antipsychotic dose reduction. Results: Existing clinical trial research is dominated by relapse as the primary outcome, with dose reduction associated with a higher risk of relapse than maintenance. Few studies have measured other patient-centered outcomes but have shown preliminary evidence for superior cognitive functioning, lower negative symptoms, and better functioning following dose reduction. Respect for autonomy is a cornerstone of psychiatric rehabilitation, and this includes the right of people to choose to reduce or discontinue antipsychotic medication. Reduced capacity for treatment decision making can be supported. Autonomy and appraisal of nonmaleficence and beneficence associated with dose reduction can be facilitated through shared or supported decision making. Clinicians should continue to strive for justice through the fair allocation of resources to support all people who request antipsychotic dose reduction. Conclusions and Implications for Practice: Clinicians have a responsibility to balance the four core ethical principles to the best of their ability when supporting a person in their recovery journey. Exploring, trialing, and supporting antipsychotic dose reduction may be part of this process if that is the patient's choice. Impact and Implications Autonomy and justice are upheld when people are supported to reduce or cease antipsychotic medication if that is their choice. Clinicians can balance the principles of nonmaleficence and beneficence (i.e., minimize harm and promote a person's welfare) by staying up-to-date with and sharing the evidence openly with patients, promoting supported or shared decision making, advanced statements, slow hyperbolic tapering, and providing additional monitoring and psychosocial support.

Large-integer extended GCD (XGCD) is a critical operation in cryptography applications such as new blockchains and modular inversion. We present the first ASIC for XGCD, which is 23x faster than state-of-the-art software, 18x faster than prior hardware simulations, and up to 303x faster than prior chips for modular inversion. These performance gains come from careful circuit and logic optimizations to avoid long carry propagation and to hide control signal delays. Our chip computes XGCD with 255-bit integer inputs in 87 ns and XGCD with 512-bit integer inputs in 176 ns on average. Our chip can be configured for constant-time execution and computes constanttime 255-bit XGCD in 119 ns and constant-time 512-bit XGCD in 239 ns.

Antidepressants are among the most extensively prescribed psychotropic drugs worldwide. Discontinuation induced withdrawal symptoms have been reported for almost all antidepressants. The incidence of antidepressant withdrawal syndrome (AWS) and other characteristics remain unknown. We searched the PubMed, Embase, PsycINFO, MEDLINE, CINAHL, and Cochrane Central Register of Controlled Trials databases from inception to December 31, 2023. Randomized double-blinded trials, longitudinal or cross-sectional studies that reported the incidence and other characteristics of antidepressant withdrawal symptoms were included. The pooled incidence of AWS was calculated by a random effects model. We included 35 studies, of which 2 studies just provided incidence of specific withdrawal symptoms, and 4 studies only described other characteristics. The pooled incidence of AWS from all available studies was 42.9%, from 11 RCTs was 44.4%, in studies in which the treatment duration was mostly 8-12 weeks, which usually appear within 2 weeks, and were generally measured for <4 weeks. The incidence in selective serotonin-norepinephrine reuptake inhibitors was the lowest (29.7%), followed by selective serotonin reuptake inhibitors (45.6%) and tricyclic antidepressants (59.7%), without significant differences (p = 0.221). Treatment duration showed a dose-response to the incidence of AWS (6-12 W: 35.1%, 12-24 W: 42.7%, >24 W: 51.4%). The half-life did not show such a simple dose-dependent relationship. The pooled estimate was robust regardless whether withdrawal symptoms were measured in RCTs or observational studies (including face-to-face and online survey studies). Tapering the dose reduced the incidence of AWS compared with abrupt stoppage (34.5% vs 42.5%), without a significant difference (p = 0.484). Risk factors for withdrawal symptoms included being female, younger, experiencing adverse effects early in treatment, taking higher doses or longer duration of medication, abrupt cessation of drugs, and those with a lower clearance of drugs or with serotonin 1A receptor gene variation. The findings suggest the incidence of AWS are common and some clinical characteristics and risk factors which can help clinicians identify who is at greater risk of experiencing AWS. Discontinuation studies on long-term antidepressant users with long follow-up periods are required in the future.

State-of-the-art deep learning models for computer vision tasks are based on the transformer architecture and often deployed in real-time applications. In this scenario, the resources available for every inference can vary, so it is useful to be able to dynamically adapt execution to trade accuracy for efficiency. To create dynamic models, we leverage the resilience of vision transformers to pruning and switch between different scaled versions of a model. Surprisingly, we find that most FLOPs are generated by convolutions, not attention. These relative FLOP counts are not a good predictor of GPU performance since GPUs have special optimizations for convolutions. Some models are fairly resilient and their model execution can be adapted without retraining, while all models achieve better accuracy with retraining alternative execution paths. These insights mean that we can leverage CNN accelerators and these alternative execution paths to enable efficient and dynamic vision transformer inference. Our analysis shows that leveraging this type of dynamic execution can lead to saving 28 % of energy with a 1.4 % accuracy drop for SegFormer (63 GFLOPs), with no additional training, and 53% of energy for ResNet-50 (4 GFLOPs) with a 3.3% accuracy drop by switching between pretrained Once-For-All models.

•Applications ranging from scientific computing to machine learning can have extremely sparse inputs

While coarse-grained reconfigurable arrays (CGRAs) have emerged as promising programmable accelerator architectures, pipelining applications running on CGRAs is required to ensure high maximum clock frequencies. Current CGRA compilers either lack pipelining techniques resulting in low performance or perform exhaustive pipelining resulting in high energy and resource consumption. We introduce Cascade, an application pipelining toolkit for CGRAs, including a CGRA application frequency model, automated pipelining techniques for CGRA application compilers that work with both dense and sparse applications, and hardware optimizations for improving application frequency. Cascade enables 7 - 34x lower critical path delays and 7 - 190x lower EDP across a variety of dense image processing and machine learning workloads, and 2 - 4.4x lower critical path delays and 1.5 - 4.2x lower EDP on sparse workloads, compared to a compiler without pipelining.

Onyx is the first fully programmable accelerator for arbitrary sparse tensor algebra kernels. Unlike prior work, it supports higher-order tensors, multiple inputs, and fusion. It achieves this with a coarse-grained reconfigurable array (CGRA) that has composable memory primitives for storing compressed any-order tensors and compute primitives that eliminate ineffectual computations in sparse expressions. Further, Onyx improves dense image processing and machine learning (ML) with application-specialized compute tiles, memory tiles optimized for affine access patterns, and hybrid clock gating in the global buffer. We achieve up to 565x better energy-delay product (EDP) for sparse kernels vs. CPUs with sparse libraries, and up to 76% and 85% lower EDP for image processing and ML, respectively, vs. Amber [1].

PURPOSE OF REVIEW:There has been an increasing focus on deprescribing in psychiatry recently, particularly of antipsychotic medication, with recognition that not all patients with psychotic disorders require lifelong medication. We summarize some empirical and theoretical papers, and examine case studies to provide instruction on this topic. RECENT FINDINGS:Recent studies have found that slower tapering (over months or longer) of antipsychotics is associated with a lower relapse rate than quicker tapering (weeks). Case studies presented suggest that the process of reduction is associated with the precipitation or exacerbation of psychotic symptoms and that a slower process of reduction may minimize this effect. This may be because faster reductions cause greater disruption of homeostatic equilibria, provoking psychotic symptoms either as direct withdrawal symptoms or consequences of nonpsychotic withdrawal symptoms (e.g. insomnia) - although not all patients will experience withdrawal symptoms. This suggests that smaller dose reductions, especially at lower doses, made very gradually, may minimize the risk of psychotic symptoms. SUMMARY:Slower tapering of antipsychotics may provide time for adaptations made to the presence of the medications to resolve, thus reducing the disruption to homeostatic equilibrium caused by dose reduction, potentially reducing the risk of relapse. Exacerbation of psychotic symptoms on antipsychotic reduction may not represent evidence of the need for a higher dose of antipsychotic on a long-term basis but may indicate the need for more gradual reduction. Gradual reduction of antipsychotics, especially after long-term use in clinical practice is prudent.