About

Shirley Wang is an Assistant Professor in the Department of Psychology at Yale University. She earned her Ph.D. in 2024 from Harvard University. Her research program aims to develop and harness novel methods that can capture and model the complexity of mental health disorders, focusing on phenomena such as suicide, nonsuicidal self-injury, and eating disorders. She integrates methods from clinical and computational sciences, including machine learning, mathematical modeling, and ambulatory assessment via smartphones and wearables. A central focus of her work is on formalizing theories of psychopathology and modeling the real-time dynamics of symptoms and their contexts as they unfold in people’s daily lives. Her ultimate goal is to use these models to inform the development of scalable, effective, and personalized interventions, such as just-in-time adaptive interventions, which aim to detect risk and intervene when support is needed.

Research topics

• Medicine
• Internal medicine
• Oncology
• Biology
• Chemistry

Selected publications

• Dynamic switching of phenylpropanoid metabolic flux mediates reversible growth-defense tradeoffs in Salix brachista

  Plant Science · 2026-05-05

  article
  PublisherDOI

• 2778P ABBV-706, a seizure-related homolog protein 6 (SEZ6)-targeting antibody-drug conjugate (ADC), in patients (pts) with relapsed/refractory (R/R) small cell lung cancer (SCLC): Circulating biomarker and molecular response analyses

  Annals of Oncology · 2025-09-01

  articleSenior author
  PublisherDOI

• Multifunctional strut-plate composite lattice metamaterial for integrated acoustic, energy, and vibration management

  Virtual and Physical Prototyping · 2025-05-11 · 11 citations

  articleOpen access

  In the pursuit of engineering solutions capable of managing persistent hazards such as noise, vibration, and structural impacts, materials that combine sound absorption, vibration damping, and deformation resistance are crucial. Lightweight lattice metamaterials have shown potentiality for these applications. These materials offer design flexibility but typically struggle to simultaneously excel in sound absorption, vibration control, and structural load-bearing. This work introduces a hollow truncated octahedron strut-plate (HTOSP) composite lattice metamaterial, employing additive manufacturing for prototype fabrication. Comprehensive validations were conducted through numerical simulations as well as impedance tube testing, with results aligning well. Adjustments in strut diameters and plate pore sizes enable the HTOSP to achieve impressive mid-high-frequency sound absorption coefficient and half-absorption bandwidth. The structural behaviour of HTOSP under load was explored through numerical simulations and quasi-static compression testing, revealing a high, stable plateau stress and specific energy absorption that peaks and then declines with increasing hollow-strut inner diameters. Furthermore, the HTOSP effectively dampens high-frequency vibrations, achieving –62 dB elastic wave attenuation at 3173 Hz via local resonance. This multifunctional HTOSP lattice metamaterial stands out for its superior performance in sound absorption, vibration control, and mechanical strength, presenting an intriguing paradigm for the design of multifunctional acoustic-mechanical-vibration structures.

  PublisherDOI

• 68eP Novel multi-biomarker nomogram combining CD155/CD226/TIGIT/CD96 immune checkpoint axis for postoperative survival prediction in breast cancer

  ESMO Real World Data and Digital Oncology · 2025-11-01

  articleOpen access
  PublisherDOI

• Catheter ablation versus medical rate control for persistent atrial fibrillation in older heart failure patients with reduced ejection fraction

  Heart · 2025-05-20

  article

  BACKGROUND: Patients with heart failure with reduced ejection fraction (HFrEF) and atrial fibrillation are mostly elderly patients, and persistent atrial fibrillation (PerAF) with multiple comorbidities tends to have a worse clinical prognosis. However, there is a lack of randomised trial to investigate the impact of catheter ablation (CA) on outcomes in older PerAF combined with HFrEF. OBJECTIVE: This study aims to compare the effects of CA versus medical rate control (MRC) on severity indicators of HFrEF. METHODS: Older patients with PerAF and HFrEF underwent transthoracic echocardiography and were randomly assigned to receive either AF ablation or MRC. The primary outcome was changes in left ventricular ejection fraction (LVEF). RESULTS: A total of 89 patients (mean age 69.5±3.9 years) were randomly allocated to the CA group (n=45) and MRC group (n=44). Baseline characteristics were similar between the two groups. After 12 months, worsening heart failure requiring unplanned hospitalisation occurred less frequently in the CA group (p=0.019). In CA group, LVEF (from baseline 36.1%±2.7% to 48.9%±7.1%; p<0.00 L) improved higher compared with the MRC group (8.7 (5.9 to 11.5)), p<0.001. Compared with baseline, New York Heart Association functional class and AF burden also showed improvement in CA group than MR group. At a follow-up period of 12 months, sinus rhythm rate was higher in CA group than MRC group, 51.1% versus 20.4%. CONCLUSION: This limited small-scale randomised study showed that CA in older patients with PerAF and HFrEF was associated with a lower likelihood of unplanned hospitalisations due to worsening heart failure with improvement in LVEF and lower AF burden. TRIAL REGISTRATION NUMBER: NCT05827172.

  PublisherDOI

• FrankenBot: Brain-Morphic Modular Orchestration for Robotic Manipulation with Vision-Language Models

  ArXiv.org · 2025-06-24

  preprintOpen access1st authorCorresponding

  Developing a general robot manipulation system capable of performing a wide range of tasks in complex, dynamic, and unstructured real-world environments has long been a challenging task. It is widely recognized that achieving human-like efficiency and robustness manipulation requires the robotic brain to integrate a comprehensive set of functions, such as task planning, policy generation, anomaly monitoring and handling, and long-term memory, achieving high-efficiency operation across all functions. Vision-Language Models (VLMs), pretrained on massive multimodal data, have acquired rich world knowledge, exhibiting exceptional scene understanding and multimodal reasoning capabilities. However, existing methods typically focus on realizing only a single function or a subset of functions within the robotic brain, without integrating them into a unified cognitive architecture. Inspired by a divide-and-conquer strategy and the architecture of the human brain, we propose FrankenBot, a VLM-driven, brain-morphic robotic manipulation framework that achieves both comprehensive functionality and high operational efficiency. Our framework includes a suite of components, decoupling a part of key functions from frequent VLM calls, striking an optimal balance between functional completeness and system efficiency. Specifically, we map task planning, policy generation, memory management, and low-level interfacing to the cortex, cerebellum, temporal lobe-hippocampus complex, and brainstem, respectively, and design efficient coordination mechanisms for the modules. We conducted comprehensive experiments in both simulation and real-world robotic environments, demonstrating that our method offers significant advantages in anomaly detection and handling, long-term memory, operational efficiency, and stability -- all without requiring any fine-tuning or retraining.

  PublisherOA PDFDOI

• The Neurophysiological Paradox of AI-Induced Frustration: A Multimodal Study of Heart Rate Variability, Affective Responses, and Creative Output

  Brain Sciences · 2025-05-25 · 2 citations

  articleOpen access

  AI code generators are increasingly used in creative contexts, offering operational efficiencies on the one hand and prompting concerns about psychological and neurophysiological strain on the other. This study employed a multimodal approach to examine the affective, autonomic, and creative consequences of AI-assisted coding in early-stage learners. Fifty-eight undergraduate design students with no formal programming experience were randomly assigned to either an AI-assisted group or a control group and engaged in a two-day generative programming task. Emotional states (PANAS), creative self-efficacy (CSES), and subjective workload (NASA-TLX) were assessed, alongside continuous monitoring of heart rate variability (HRV; RMSSD and LF/HF). Compared to the controls, the AI-assisted group exhibited greater increases in negative affect (p = 0.006), reduced parasympathetic activity during the task (p = 0.001), and significant post-task declines in creative self-efficacy (p &lt; 0.05). Expert evaluation of creative outputs revealed a significantly lower performance in the AI group (p = 0.040), corroborated by behavioral observations showing higher tool dependency, emotional volatility, and rigid problem-solving strategies. These findings indicate that, in novice users, the opacity and unpredictability of AI feedback may disrupt emotional regulation and autonomic balance, thereby undermining creative engagement. The results highlight the need to consider neurocognitive vulnerability and the learner’s developmental stage when integrating AI tools into cognitively demanding creative workflows.

  PublisherOA PDFDOI

• AntiGrounding: Lifting Robotic Actions into VLM Representation Space for Decision Making

  ArXiv.org · 2025-06-14

  preprintOpen access

  Vision-Language Models (VLMs) encode knowledge and reasoning capabilities for robotic manipulation within high-dimensional representation spaces. However, current approaches often project them into compressed intermediate representations, discarding important task-specific information such as fine-grained spatial or semantic details. To address this, we propose AntiGrounding, a new framework that reverses the instruction grounding process. It lifts candidate actions directly into the VLM representation space, renders trajectories from multiple views, and uses structured visual question answering for instruction-based decision making. This enables zero-shot synthesis of optimal closed-loop robot trajectories for new tasks. We also propose an offline policy refinement module that leverages past experience to enhance long-term performance. Experiments in both simulation and real-world environments show that our method outperforms baselines across diverse robotic manipulation tasks.

  PublisherOA PDFDOI

• Heterochronic parabiosis alters the transcriptomic landscape to combat aging and aging-related diseases in aging-accelerated mice

  Life Medicine · 2025-06-29 · 2 citations

  articleOpen access

  Aging is a multifactorial process involving a gradual decline in cellular and tissue functions, making it a major risk factor for aging-related degenerative diseases. In this study, we utilized the senescence-accelerated mouse prone 8 mice model, which mimics pathological characteristics of Alzheimer's disease, fatty liver disease, and cardiac fibrosis, to construct a heterochronic parabiosis model and systematically investigate the rejuvenating effects of heterochronic parabiosis on the brain, liver, and heart. Our findings revealed that heterochronic parabiosis promotes synaptic plasticity and neuronal communication, restores hepatocyte metabolic functions, and reduces chronic inflammation and fibrosis in the heart. Notably, heterochronic parabiosis significantly downregulates the expression of age-related disease risk genes. In addition, endothelial cells, as cell types directly exposed to the circulatory environment, demonstrated the highest sensitivity to heterochronic parabiosis across three organs, and exhibited significantly reduced inflammation after intervention, suggesting that they may play an early and central role in the rejuvenation process. Overall, our study increases the understanding of the molecular and cellular mechanisms of aging and its related diseases, highlights the multiorgan and multitarget potential of heterochronic parabiosis in delaying aging and mitigating aging-related diseases, and provides new therapeutic targets for achieving healthy aging.

  PublisherOA PDFDOI

• 0315 Caution regarding historical estimates and trends for global burden of AD

  Journal of Investigative Dermatology · 2025-07-21

  articleOpen access1st authorCorresponding
  PublisherOA PDFDOI

Recent grants

• Development and Evaluation of a Decision Analysis Based Decision Aid

  NIH · $754k · 2015–2020

• Comparative modeling of multiple myeloma across myeloma continuum: prevention, treatment, and health drivers

  NIH · $664k · 2021–2026

• Comparative modeling of multiple myeloma across myeloma continuum: prevention, treatment, and health drivers

  NIH · $2.5M · 2021–2026

• Project 1: Targeting Tumor Phospholipase PLA2G10 for Cancer Immunotherapy

  NIH · $49.5M · 2025–2030

Frequent coauthors

• Cary P. Gross

  173 shared

• Brigid K. Killelea

  Brigham and Women's Hospital

  62 shared

• James B. Yu

  Yale Cancer Center

  48 shared

• Suzanne B. Evans

  Yale University

  48 shared

• Natalia Kunst

  University of York

  46 shared

• Craig Evan Pollack

  University of Baltimore

  45 shared

• Sarah S. Mougalian

  Yale University

  42 shared

• Pamela R. Soulos

  Yale Cancer Center

  42 shared

Education

• PhD, Health Policy and Management

  University of Minnesota System

  2012

• MD, Medicine

  Taipei Medical University

  1992