The researchers argue that most existing approaches to anxiety detection suffer from a fundamental conceptual flaw: they treat anxiety as a discrete category, a yes-or-no label, rather than as a mental state that ebbs, flows, and varies in intensity over time. Convolutional and recurrent neural networks, the workhorses of physiological signal analysis, also struggle to incorporate uncertainty, inter-subject variability, and the hidden hierarchy of emotional severity. And crucially, most systems stop at prediction. They tell you that someone seems anxious, but offer no guidance on what to do about it. DLAQ-ERRG was designed to close that gap with a single end-to-end pipeline that moves from raw sensor data to actionable, explainable support.

The architecture rests on three interlocking components. The first, a Neural Stochastic Differential Attention Encoder (NSDAE), models the latent anxiety state as a continuous-time stochastic process governed by a stochastic differential equation. Instead of chopping biosignals into static windows and treating each one independently, the encoder learns a drift function, parameterized by an attention mechanism that dynamically weights the contribution of each physiological modality, and a diffusion term that captures random fluctuations, sensor noise, and individual differences. Amortized variational inference with a Gaussian posterior allows the model to output not just a point estimate of arousal but a calibrated confidence interval around it, distinguishing sudden emotional reactions from enduring patterns.

The second component, a Hyperbolic Contrastive Affective Manifold Learner (HCAML), addresses a subtler representational problem. Mild, moderate, and severe states of arousal form a hierarchical, tree-like structure, and flat Euclidean latent spaces handle such geometry poorly. HCAML projects the learned representations onto a Poincaré ball with fixed curvature of minus 1.0, where distances expand exponentially toward the boundary of the manifold. In this curved space, low-arousal states cluster near the center while higher-severity states spread toward the periphery, and a radial ordering regularization term enforces that greater severity corresponds to greater distance from the origin. A contrastive loss with an InfoNCE formulation, using a temperature of 0.07, pulls temporally adjacent windows from the same subject and condition together while pushing apart samples from different subjects and conditions.

The third component, an Entropy-Regularized Neuro-Symbolic Policy Generator (ER-NSPG), turns quantified arousal into decisions. The action space is deliberately small and human-readable: paced diaphragmatic breathing at six to eight breaths per minute, guided mindfulness grounding, cognitive reappraisal prompts, a brief physical pause or stretch, a social support suggestion, or passive monitoring. Symbolic rules derived from cognitive-behavioral intervention logic constrain which actions are admissible at a given arousal level. When predicted intensity exceeds a high threshold, only physiological down-regulation strategies such as paced breathing and grounding are permitted; moderate states allow cognitive reappraisal; low-intensity states default to monitoring. A reinforcement learning policy trained with proximal policy optimization, entropy regularization, and temporal smoothness penalties selects among the admissible actions, preventing both overconfident determinism and erratic switching between interventions.

Training and evaluation relied on the WESAD dataset, a public benchmark of wearable stress and affect detection containing chest- and wrist-worn sensor recordings from 15 laboratory volunteers, including electrocardiography, electrodermal activity, respiration, and blood volume pulse. Because WESAD lacks continuous anxiety annotations, the team constructed a proxy target: a normalized arousal index blending experimentally induced stress activation with min-max-normalized self-reported stress ratings, weighted 0.7 and 0.3 respectively. The authors are careful to stress that this index represents anxiety-related physiological arousal in an affective computing sense, not a clinical diagnosis of an anxiety disorder. Validation used a strict Leave-One-Subject-Out protocol, in which each of the 15 subjects was held out in turn, with all preprocessing parameters estimated exclusively from training subjects to eliminate data leakage.

The results are striking. The full framework achieved a Concordance Correlation Coefficient of 0.921 plus or minus 0.028, a Mean Absolute Error of 0.061, and a Root Mean Square Error of 0.083, with classification accuracy of 97.3 percent for the three-class affect task. Compared against the strongest baseline, a Transformer encoder scoring 0.882, the improvement was statistically significant, with a paired t-test yielding p equal to 0.0003 and a large effect size of 1.26. Ablation studies confirmed that each component earns its place: replacing the stochastic encoder with a BiLSTM dropped the CCC to 0.889, while swapping hyperbolic embeddings for Euclidean ones of identical parameter count reduced the Anxiety Severity Separation Score by 6.8 percent. Removing entropy regularization cut recommendation diversity by 11.2 percent, and removing smoothness regularization increased temporal action variance by 14.7 percent.

The recommendation module was evaluated on multiple fronts. It scored 0.776 on the Action Appropriateness Score, 0.764 on the Recommendation Smoothness Score, and 0.701 on the Emotion Regulation Diversity Score. A subset of 120 state-action pairs was independently rated by three domain-informed evaluators, reaching a Cohen’s kappa of 0.81, and the appropriateness score correlated with mean expert ratings at r equal to 0.74. Case studies illustrate the system in action: for one subject in a high-arousal stress phase, predicted intensity of 0.81 triggered three consecutive windows of paced breathing, after which predicted intensity fell to 0.63 within 15 seconds and the uncertainty band narrowed. In a failure case involving motion artifacts, the model’s prediction briefly spiked while ground truth held steady, but the elevated uncertainty flagged the problem and the system reverted to passive monitoring once signal variance dropped.

Calibration, often the Achilles heel of probabilistic models, proved strong here. The Expected Calibration Error was just 0.027 across folds, and the correlation between predicted standard deviation and absolute regression error was 0.61, meaning the model genuinely knows when it is unsure. Discarding the top 10 percent of highest-uncertainty samples reduced RMSE from 0.083 to 0.071, demonstrating that the uncertainty estimates could support reliability-aware deployment. On the computational side, the full model contains roughly 2.31 million parameters and runs a forward pass in 18.6 milliseconds on a GPU or 74.3 milliseconds on an ordinary Intel i7-class CPU, comfortably fast for real-time wearable use. A pruned wrist-only configuration with a reduced latent dimension cuts inference latency to under 50 milliseconds with a negligible accuracy penalty.

The authors are candid about the limits. WESAD contains only 15 subjects recorded under controlled laboratory conditions with an acute, experimentally induced stress protocol, so generalization to chronic anxiety, clinical populations, and real-world environments remains unproven. The symbolic rules are fixed rather than adapted to individual coping styles or cultural context, and the recommendations have not been validated as therapeutic interventions in outcome trials. A small pilot study with 18 participants found high ratings for clarity and appropriateness of the explanations, but noted that overly frequent recommendations hurt perceived usability. Privacy of sensitive physiological data, fairness across demographics, and the need for human oversight all receive explicit attention. Still, as a proof of concept, the work points toward a future in which a smartwatch does not merely count your steps, but quietly tracks your autonomic state, tells you how it knows, and suggests a breathing exercise at exactly the moment it might help.

Subject of Research: Deep learning-based continuous quantification of anxiety-related physiological arousal and generation of personalized emotional regulation recommendations from multimodal wearable signals

Article Title: Deep learning-based anxiety quantification assessment and emotional regulation recommendation generation model

Article References: Zhai, J., & Wan, Q. (2026). Deep learning-based anxiety quantification assessment and emotional regulation recommendation generation model. Discover Artificial Intelligence, 6(1), Article 1357. https://doi.org/10.1007/s44163-026-01886-w

Image Credits: AI Generated

DOI: 10.1007/s44163-026-01886-w

Keywords: anxiety quantification, affective computing, deep learning, wearable sensors, stochastic differential equations, hyperbolic representation learning, neuro-symbolic reasoning, emotion regulation, uncertainty-aware modeling, WESAD dataset, reinforcement learning, digital mental health

The story begins with the Sharma-Parthasarathy stochastic two-body problem, introduced in 2007 to describe a particle moving through a stochastically fluctuating dust cloud. In that model, the familiar deterministic equations of motion are augmented by white-noise terms with separate strengths in the radial and angular directions. The mathematical consequence is dramatic. In the deterministic Kepler problem, energy and angular momentum are strong first integrals, quantities that remain exactly constant along every trajectory. In the stochastic version, that guarantee evaporates. Previous work by Cresson, Pierret and Puig established that the angular momentum loses its strong status but survives as a weak integral, meaning its statistical expectation stays constant in time even as individual trajectories scatter. The energy, by contrast, fails to qualify even as a weak integral: on average, noisy orbits steadily gain or lose energy.

Celletti and Lhotka add a second physical ingredient that is impossible to ignore in dust-cloud environments: dissipation. They model it as a generalized Stokes-type drag force, following the classical framework of Breiter and Jackson, in which friction strengths differ between the radial and transverse directions. This family includes three famous special cases from the history of celestial mechanics: the Poynting-Plummer-Danby drag, where radial and transverse coefficients are equal; the Poynting-Robertson drag, familiar from studies of dust grains spiraling into the Sun, where the radial coefficient is twice the transverse one; and the Poynting-Plummer case, with a ratio of four. Alone, dissipation drains both energy and angular momentum from an orbit, contracting it and circularizing its path. The authors prove analytically that under pure dissipation a modified angular momentum, the ordinary one plus a term proportional to the drag coefficient and the accumulated orbital angle, is exactly preserved, a result that traces its lineage back to Poynting’s 1904 work on radiation in the solar system.

The heart of the new paper is the question of what happens when both effects act together. Noise alone pumps energy into the orbit on average; drag alone bleeds it away. Could there be parameter values, or particular initial conditions, for which these opposing tendencies cancel so precisely that the energy behaves once more as a weak integral? Celletti and Lhotka answer with an explicit theorem. The modified angular momentum, which folds the dissipative correction into its definition, is always a weak integral of the combined stochastic-dissipative dynamics, regardless of the parameters chosen. The total energy, meanwhile, becomes a weak integral if and only if a balance condition holds: the expected cumulative energy injected by the stochastic fluctuations, weighted by the radial and angular noise strengths, must exactly equal the expected cumulative energy removed by the radial and transverse drag components. Written out, the condition links the noise amplitudes, the drag coefficients, and time-averaged measures of the orbital velocities in a single elegant equation.

The proof itself is a compact exercise in stochastic calculus. Using the multi-dimensional Ito formula, the authors compute the differential of the energy along solutions of the combined stochastic-dissipative equations, separating deterministic drift terms from purely stochastic ones. The stochastic pieces, being integrals against Brownian motions, vanish in expectation. What remains is a deterministic expression for the mean energy drift, which vanishes exactly when the balance condition is satisfied. A parallel computation for the modified angular momentum shows that its stochastic differential contains only a noise term with zero expectation, guaranteeing its weak conservation without any parameter constraint at all. The Itô and Stratonovich formulations of the problem coincide, since the Wong-Zakai correction term vanishes, so the results hold in either interpretation of the stochastic calculus.

To show that the theorem is more than an abstract curiosity, the authors turn to concrete orbital examples. They begin with a particle on an ellipse with semimajor axis one, eccentricity 0.05, and unit mean motion, expanding the orbital variables in Taylor series of the small eccentricity. From these expansions they derive an explicit formula for the critical drag parameter that satisfies the energy balance for given noise strengths. For radial and angular volatilities of 0.01 and 0.002 respectively, the critical value comes out to approximately 0.0000518622. They then integrate the full stochastic-dissipative equations numerically, propagating one thousand independent paths for each of three drag choices: five percent below the critical value, exactly at it, and five percent above. The results are textbook-clean. Below the critical drag, the mean energy climbs; above it, the mean energy decays; at the critical value, the averaged energy oscillates around its initial level with no secular trend, confirming the weak-integral property within the accuracy of the approximation.

The numerical campaign also exposes the limits of the theory. Because energy fixes the semimajor axis and angular momentum fixes the eccentricity through the classical Keplerian relations, any drift in the angular momentum necessarily inflates the eccentricity, and as the eccentricity grows the balance condition, derived at the initial orbital elements, gradually loses its validity. Tracking the osculating elements along each trajectory, the authors quantify the error introduced by holding the elements fixed, finding correction terms proportional to products of the drag coefficients with the squared rates of change of the orbital elements. Over longer integrations, reaching a full thousand years, the averaged orbit initially holds its elliptical shape but eventually spirals outward, with the mean eccentricity racing past 0.2 and most paths escaping to unbound, hyperbolic trajectories. The balance, in other words, is real but local in time, a delicate equilibrium that orbital evolution itself eventually destroys.

The authors also enrich the noise model in a physically motivated direction. Justifying the white-noise idealization through the central limit theorem, since the gravitational tug of countless individual dust particles superposes into an approximately Gaussian random force, they extend the stochastic forcing to general Wiener processes with nonzero drift. The drift terms can be interpreted as slow changes in the density of the perturbing dust cloud. Remarkably, a suitable adjustment of these drifts can improve the conservation of the mean energy, playing a role analogous to the dissipation coefficients themselves. In sample simulations, four different drift choices show that the drift acts as an additional knob for tuning the secular energy trend, hinting at practical control strategies for maintaining quasi-conservative average dynamics in noisy, dissipative environments.

The implications reach beyond idealized dust clouds. Any small body embedded in a gaseous or particulate medium, from circumstellar dust grains to space debris in tenuous atmospheres, experiences some combination of stochastic perturbation and drag, and the question of which quantities survive on average is central to long-term orbit prediction. By proving that energy conservation can be restored, on average, by a precise match between noise intensity and dissipation strength, and by identifying a modified angular momentum that is always weakly conserved, Celletti and Lhotka provide both a conceptual framework and explicit computational recipes for that task. The authors regard their results as a promising foundation for extending the analysis to richer dynamical settings, including the notoriously chaotic three-body problem and rotational models such as the spin-orbit coupling problem, where the interplay of randomness and dissipation remains largely unexplored. For now, the humble two-body problem has yielded one more surprise: even under noise and friction, order can re-emerge, provided the chaos and the damping are tuned to cancel each other exactly.

Subject of Research: The stochastic two-body problem with dissipative Stokes drag and the conditions under which energy and angular momentum act as weak integrals of motion.

Article Title: The stochastic two-body problem with dissipation

Article References: The stochastic two-body problem with dissipation. (n.d.). https://doi.org/10.1007/s10569-026-10329-8

Image Credits: AI Generated

DOI: 10.1007/s10569-026-10329-8

Keywords: two-body problem, stochastic dynamics, dissipation, Stokes drag, weak integrals, angular momentum, orbital energy, white noise, dust clouds, celestial mechanics, Poynting-Robertson drag, stochastic differential equations