IntroductionAdolescent depression has emerged as a critical global public health concern, with rising prevalence in China posing severe threats to psychological development and social adaptation. Traditional statistical methods face limitations in capturing complex non-linear relationships and interactions among influencing factors, while machine learning algorithms offer advantages in predictive modeling of mental health disorders.ObjectiveThis study aimed to: (1) compare the performance of seven ML algorithms in classifying low and high depression risk groups among Chinese adolescents; (2) identify key predictive factors from demographic, personality, and PGI-related variables; (3) explore non-linear relationships and interactive effects between critical factors; and (4) explore preliminary threshold values for key factors as potential references for risk identification.MethodsA total of 559 Chinese adolescents completed assessments of demographic characteristics, Big Five personality traits, personal growth initiative, and depression symptoms. Model performance was compared using Friedman tests and Nemenyi post-hoc tests appropriate for correlated cross-validation data. Seven ML algorithms were trained and optimized using 5-fold cross-validation. Feature importance was analyzed via traditional metrics and SHAP values, and SHAP interaction effects were tested using permutation tests. Threshold analysis was conducted using the Youden’s J statistic.ResultsLightGBM outperformed other models with an AUC of 0.834, achieving balanced accuracy, sensitivity, and specificity. Neuroticism emerged as the most robust predictor across all models, followed by proactive change, agreeableness, extraversion, and growth resilience. Demographic factors showed minimal predictive power. SHAP permutation tests confirmed significant interactions between neuroticism and proactive change and between proactive change and agreeableness, whereas no significant interaction was found between neuroticism and agreeableness. Preliminary thresholds were identified for key factors within this sample.ConclusionML algorithms, particularly lightGBM, effectively identify adolescent depression risk, with personality traits and PGI serving as core predictive factors. The findings highlight the value of integrating multi-dimensional variables in depression prediction and provide preliminary references for early intervention. Given the cross-sectional design and lack of external validation, conclusions regarding generalizable cutoffs and causal inference should be made with caution. Targeted strategies focusing on reducing neuroticism and enhancing proactive growth behaviors may mitigate depression vulnerability in Chinese adolescents.
Understanding the workplace needs of autistic adults in Singapore: insights to inform inclusive AI support
IntroductionAutistic adults face persistent challenges in obtaining and sustaining meaningful employment. Despite growing attention to workplace inclusion, research on autistic adults’ employment experiences in non-Western contexts remains scarce. The potential of emerging technologies, such as large language models (LLMs), to support workplace integration is also still largely unexplored, yet off-the-shelf models may reflect neurotypical norms that risk reinforcing masking or overlooking neurodivergent needs. This qualitative, participatory study therefore centered autistic adults’ perspectives to understand workplace experiences and identify potential LLM affordances that could facilitate integration.MethodsTwenty autistic adults with at least 3 months of work experience were recruited in Singapore. Data was collected through semi-structured group discussions. Using reflexive thematic analysis, we identified workplace challenges, needs, and potential support LLMs can provide. We had a multidisciplinary team of autistic and non-autistic researchers, and autistic perspectives actively shaped the design, conduct, and interpretation of the research.ResultsTwo overarching themes emerged: (1) assumed neurotypicality of the workplace, evident in work processes and social participation, and (2) need for workplace inclusivity, supported through both individual accommodations and systemic change, including identity-affirming support, alignment of work design and tools with neurodivergent working styles, empowered access to supports and accommodations, and shared responsibility for workplace integration. Potential LLM functionalities involve supporting executive functioning, encouraging self-reflection, and fostering mutual understanding between autistic employees and their coworkers.DiscussionWorkplace barriers for autistic employees often stem from assumed neurotypical norms rather than individual deficits. Participants reported challenges related to ambiguous work processes, implicit social expectations, and executive functioning demands, which reflect a mismatch between workplace structures and neurodivergent ways of working. Crucially, inclusivity cannot rely solely on individual accommodations; meaningful workplace inclusion requires systemic change. Designing LLM tools that align with neurodivergent working styles can complement systemic inclusivity efforts and empower autistic employees. Implications and future directions are discussed.
T cell-engaging bispecific antibodies for myeloid malignancies: Targets, formats, and clinical challenges
Paczesny et al. review why bispecific antibodies, successful in other blood cancers, have lagged in acute myeloid leukemia. They highlight how limited tumor specificity and toxicity hinder progress and outline emerging targets and engineering strategies that could enable safer, more effective immunotherapies for this hard-to-treat disease.
Antibody-drug conjugates in breast cancer: Progress and future directions
Newman et al. review the evolving landscape of antibody-drug conjugates (ADCs) in breast cancer, including approved agents, resistance mechanisms, and combinatorial strategies. The authors highlight novel agents, antigen targets, and next-generation platforms, underscoring the need for predictive biomarkers and optimized sequencing strategies to improve patient selection and efficacy.
Microbiome biomarkers in autism spectrum disorder: Toward prediction, diagnosis, and prognosis
Wong et al. review gut microbiota biomarkers for ASD, detailing gut-brain axis mechanisms, biomarker evolution from taxonomic profiling to multi-omics and multi-kingdom integration, key translational challenges, and actionable research guidelines to boost clinical utility for ASD diagnosis and prognosis.
BDNF insufficiency exacerbates ALS progression
Xu and He et al. provide in vivo evidence that BDNF insufficiency exacerbates the symptoms and pathologies in amyotrophic lateral sclerosis (ALS) patients and mouse models, leading to accelerated demise. They develop an agonistic antibody that boosts BDNF signaling and demonstrates therapeutic benefits in multiple models of ALS mice.
A human iPSC-derived sensory neuron platform for high-throughput discovery of neuroprotectants against chemotherapy-induced peripheral neuropathy
Petrova et al. develop a high-throughput human iPSC-derived sensory neuron platform for drug discovery in chemotherapy-induced peripheral neuropathy. They uncover that inhibition of three members of the STE20 kinase family (MAP4K4/MINK1/TNIK) is neuroprotective against paclitaxel-induced axon damage, with validation in primary human and mouse models.
Bridging the gap in TB case detection
Nature Medicine, Published online: 19 May 2026; doi:10.1038/d41591-026-00026-2
The portable, low-resource MiniDock MTB test, evaluated across multiple countries, shows promise for detecting pulmonary tuberculosis using sputum and tongue swabs.
STAT+: Marc Tessier-Lavigne addresses new book’s allegations about his conduct in Stanford misconduct case
SAN FRANCISCO — Former Stanford President Marc Tessier-Lavigne responded publicly for the first time Tuesday to allegations in a new book that he was forced to resign from the university not only because of flaws in his oversight of scientists but over how he handled the controversy.
At the STAT Breakthrough Summit West, STAT reporter Matthew Herper read aloud three paragraphs from Theo Baker’s book, “How to Rule the World.’’ Tessier-Lavigne sat with hands clasped in his lap as he listened to Baker’s description of the board meeting that led to his ouster.
According to Baker, the board concluded “that Tessier-Lavigne’s admit-nothing, deny-everything approach ‘did not reflect well on him and, by extension, the institution.’” The Stanford investigation, according to unnamed sources in the book, omitted yet another incident that contributed to the university turning on Tessier-Lavigne — a younger, female colleague challenging the conclusions of his work, and him dismissing her. By the end of the meeting, “there was no pro-MTL camp” and the board voted unanimously to replace him, Baker wrote.
Oncogenic Signaling Shaped by a Golgi Trafficking Protein Pair
A new study in Science Signaling identifies a previously overlooked control point in receptor tyrosine kinase (RTK) signaling, one that operates not at the plasma membrane, but at the Golgi. The research, published as “Oncogenic receptor tyrosine kinase signaling is driven by the Golgi protein GOLPH3 and its interaction with MYO18A,” reveals that the Golgi‑localized proteins GOLPH3 and MYO18A act together to route RTKs to the cell surface, thereby setting the strength of growth‑factor signaling across multiple pathways.
The work was led by Kyle Starost and colleagues at Case Western Reserve University School of Medicine and the University of California, San Diego. Their findings help explain why GOLPH3 is frequently amplified in human cancers and why its overexpression correlates with poor prognosis across tumor types.
RTKs such as EGFR, insulin receptor, and PDGFR are central drivers of proliferation and survival in many cancers. Although RTK inhibitors are widely used clinically, resistance often emerges, underscoring the need for alternative strategies that modulate signaling upstream of the receptor. The new study identifies one such upstream node: the delivery of RTKs from the Golgi to the plasma membrane.
Using an unbiased signaling analysis, the team found that siRNA knockdown of GOLPH3 or MYO18A impaired phosphorylation of EGFR at Tyr1068 and Tyr1086, as well as downstream AKT and ERK signaling. These defects persisted even when PI3K/AKT/mTOR signaling was pharmacologically blocked, demonstrating that GOLPH3 acts directly at the receptor level rather than through mTOR modulation.
To pinpoint the mechanism, the researchers turned to trafficking assays. Imaging of endogenous EGFR showed that loss of GOLPH3 or MYO18A caused the receptor to accumulate in intracellular puncta rather than at the plasma membrane. A quantitative PDGFR‑GFP surface‑delivery assay confirmed that both proteins are required for Golgi‑to‑surface transport. Treatment with brefeldin A or golgicide A, which disrupt Golgi structure, produced similar reductions in surface receptor levels, reinforcing the conclusion that the GOLPH3–MYO18A complex is essential for RTK delivery.
Overexpression experiments completed the mechanistic picture. Increasing GOLPH3 or MYO18A levels enhanced EGF‑stimulated phosphorylation of EGFR and AKT, while a GOLPH3 mutant unable to bind PI4P failed to do so. These results position the GOLPH3–MYO18A complex as a central determinant of RTK availability at the cell surface.
The authors wrote, “The GOLPH3-MYO18A complex at the Golgi apparatus was required and rate-limiting for RTK signaling across the cell types and receptors assessed.” The findings suggest that targeting Golgi‑based trafficking machinery could offer a new therapeutic angle for tumors that rely on hyperactive RTK signaling or have developed resistance to RTK inhibitors.
The post Oncogenic Signaling Shaped by a Golgi Trafficking Protein Pair appeared first on GEN – Genetic Engineering and Biotechnology News.