Hongyu Wang

Dr. Hongyu Wang

orcid

Research Focus & Metrics

Academic Career Length

1Year (2026-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
11 August 2026 in Electronics
Verified
Journal Article
Open access
PDF file
HCHaoze Chen
Haoze Chen
HWHongyu Wang
Hongyu Wang
XGXiang Gao
Xiang Gao
FZFan Zhou
Fan Zhou
Scalp electroencephalography (EEG) could support low-burden monitoring of Parkinson’s disease (PD), but participant leakage, electrode count, and dataset shift can overstate performance. We tested whether a Gumbel selector could reduce eight candidate inputs to four without using outer test data, then retrained every setting with the same backend, which fused a bidirectional long short-term memory multiple instance learning (LSTM-MIL) temporal branch with a participant-level spectral random forest. OpenNeuro ds004584 included 144 participants (97 PD and 47 controls) and 7000 four-second epochs evaluated by five-fold participant-wise cross-validation across five seeds. Among the four primary fused input settings, Full8 had the highest mean balanced accuracy (BA), area under the receiver operating characteristic curve (AUC), and F1 score (0.6698 ± 0.0337, 0.7349 ± 0.0110, and 0.7431 ± 0.0166), while Learned4 achieved BA 0.6596 ± 0.0502 compared with 0.6338 ± 0.0450 for Fixed4 and 0.6367 ± 0.0202 for Random4; exact paired tests found no clear ranking. In event-defined eye-open ds003490 (25 PD-OFF and 25 controls), a separate eight-input benchmark trained within the cohort under Fixed19, using LR/ET spectral models and fixed fusion and decision rules, had the highest observed BA of 0.5760 (95% CI, 0.4520–0.6960). Under Fixed19, direct application of the frozen Learned4 source models yielded BA 0.5592 (95% CI, 0.5184–0.6024) and AUC 0.6417 (0.5460–0.7355). Learned4 produced 18 input sets across 25 combinations of fold and seed, and every paired transfer contrast with the other settings included zero. Taken together, Learned4 achieved a source cohort mean BA 0.0102 below Full8 and showed modest discrimination when the frozen source models were transferred under Fixed19, without a clear paired advantage over the controls. Because both datasets used 16-channel average referencing, their quality control procedures differed, and the selected subsets were unstable; acquisition-compatible validation with a predefined montage is needed before sparse model inputs can inform practical electrode reduction.

4 September 2026 in SSRN Electronic Journal
Verified
Preprint
Open access
YSYuting Sun
Yuting Sun
HWHongyu Wang
Hongyu Wang
HNHaoqi Ni
Haoqi Ni
YYYufang Yang
Yufang Yang
XGXiang Gao
Xiang Gao
KXKedi Xu
Kedi Xu
Background: Early seizure prediction could enable timely intervention, but the electrophysiological organisation of the preictal state remains incompletely characterised. Phase–amplitude coupling (PAC) captures interactions between slower rhythms and faster oscillations, but prior studies have focused on ictal detection and epileptogenic tissue localisation, leaving preictal PAC dynamics insufficiently characterised. Methods: We represented PAC using its complex form and developed AdaST-Net for seizure prediction. AdaST-Net combines adaptive channel fusion for variable electrode configurations, ConvNeXt-V2 and SimAM for spatial–spectral feature extraction, and bidirectional Mamba for temporal modelling. Performance was evaluated on CHB-MIT, Kaggle canine intracranial EEG, and cTLER rat LFP datasets. Findings: Coupling strength increased during the preictal period in preictal across datasets, with modest theta-related preferred-phase shifts towards the low-frequency peak in CHB-MIT and cTLER. Subject-dependent sensitivities were 88.30 %, 96.15 %, and 94.03 %. Event-based sensitivities were 89.23 % for CHB-MIT and 97.70 % for cTLER, with lead times of 46.14 min and 30.54 min. Cross-dataset sensitivity remained high, although transfer to CHB-MIT reduced balanced accuracy. Imaginary-component occlusion and magnitude-preserving phase randomisation impaired performance. Theta–low gamma and theta–gamma PAC contributed consistently. Information was distributed across frequency pairs, whereas fewer channels retained comparable performance. Interpretation: Complex PAC involves changes in both coupling strength and the phase organisation of cross-frequency activity in preictal state, serving not only as an effective biomarker for seizure prediction but also providing a new perspective for analysing multidimensional and multiscale brain networks.

4 July 2026 in Biology
Verified
Journal Article
Open access
PDF file
THTingting He
Tingting He
HWHongyu Wang
Hongyu Wang
HNHaoqi Ni
Haoqi Ni
YSYuting Sun Yuting Sun
Yuting Sun Yuting Sun
XGXiang Gao
Xiang Gao
KXKedi Xu
Kedi Xu
Levodopa-induced dyskinesia (LID) is a severe motor complication associated with long-term levodopa (L-DOPA) treatment for Parkinson’s disease (PD). Its underlying mechanisms remain unclear, and candidate biomarkers lack consistency. To investigate cortico-striatal network alterations associated with LID, we simultaneously recorded single-neuron spikes and local field potentials (LFPs) from the dorsolateral striatum (DLS) and the primary motor cortex (M1) in LID rats. Our results showed that in the DLS, the LID group had a greater number of putative fast-spiking interneurons (FSIs) with lower firing rates, and fewer putative medium spiny neurons (MSNs) with higher firing rates. In M1, pyramidal neurons were fewer but fired faster, while interneurons were more numerous with no change in firing rate. Although gamma power increased and delta power decreased in both regions in LID rats, delta-gamma phase-amplitude coupling (PAC) was present in the DLS but absent in M1. Furthermore, cross-regional PAC analysis revealed significantly stronger coupling between the low-frequency phase of M1 and the high-frequency amplitude of the DLS than in the opposite direction, indicating an asymmetric pattern of cortico-striatal coupling in LID. These findings demonstrate region-specific alterations in neuronal activity and oscillatory coupling associated with LID and suggest that asymmetric cortico-striatal PAC may serve as a promising electrophysiological marker for characterizing abnormal network dynamics underlying dyskinesia.