A hypergraph neural network-based framework for identification of Parkinson’s disease subtypes

doi:10.1088/1674-1056/ae2677

Abstract Parkinson's disease (PD) exhibits significant phenotypic heterogeneity, which complicates clinical management and underscores the need for precise subtyping. Existing subtyping approaches often rely on a single modality, such as clinical assessments, failing to capture the complex, multi-faceted nature of the disease. This paper proposes a novel computational framework that integrates multi-modal data, specifically preprocessed functional MRI, DNA methylation, and clinical behavioral assessments, for PD subtyping. The methodology involves constructing individual hypergraphs for each modality using $K$-nearest neighbors (KNN), followed by the integration of these hypergraphs into a unified, multi-modal hypergraph using similarity network fusion (SNF). This consolidated hypergraph is then processed via a hypergraph neural network (HGNN) utilizing hyperedge convolution to cluster patients into distinct subtypes. Our experimental results demonstrate that this approach effectively identifies PD subtypes with significant clinical and biological relevance. We provide a comprehensive analysis of the model's performance and further validate the reliability of the identified subtypes through post-hoc statistical tests. This study highlights the potential of graph-based machine learning in disentangling disease heterogeneity, paving the way for personalized therapeutic strategies and improved patient outcomes.

Keywords: hypergraph neural network similarity network fusion Parkinson's disease subtypes multi-modal data

Received: 07 November 2025
Revised: 20 November 2025
Accepted manuscript online: 02 December 2025

PACS:	89.75.Hc	(Networks and genealogical trees)
	87.19.x
	87.19.La
	89.70.-a	(Information and communication theory)

Corresponding Authors: Meili Lu
E-mail: meililu@tute.edu.cn

Cite this article:

Meili Lu(卢梅丽) and Hongbao Lu(卢宏保) A hypergraph neural network-based framework for identification of Parkinson’s disease subtypes 2026 Chin. Phys. B 35 078703

[1] Venkateshwarla R R 2021 IP Indian J. Neurosci. 7 148
[2] Kandharkar S S, Kashid N A, Kanase S B and Mane D 2021 undefined
[3] Heimrich K G, Schönenberg A, Santos-García D, Mir P and Prell T 2023 J. Clin. Med. 12 2573
[4] Deuschl G and Krack P 2020 Lancet Neurol. 19 472
[5] Storey K, Foster E, Kalishman S and Doty T 2021 Arch. Phys. Med. Rehabil. 102 e37
[6] Pérez-Lloret S, Chevalier G, Bordet S, Barbar H, Capani F, Udovin L and Otero-Losada M 2023 undefined
[7] Wolters A F, Michielse S, Kuijf M L, Defebvre L, Lopes R, Dujardin K and Leentjens A F G 2022 Parkinsonism Relat. Disord. 105 32
[8] Markello R D, Shafiei G, Tremblay C, Postuma R B, Dagher A and Mišić B 2021 npj Parkinsons Dis. 7 6
[9] Yakovenko E V, Fedotova E Y and Illarioshkin S N 2020 Ann. Clin. Exp. Neurol. 14 56
[10] Lin Q, Ding H, Zhang Z, Gu Z, Ma J, Chen L, Chan P and Cai Y 2012 Neurosci. Lett. 507 147
[11] Shah A, Nguyen T A K, Peterman K, Khawaldeh S, Debove I, Shah S A, Torrecillos F, Tan H, Pogosyan A, Lachenmayer M L, Michelis J P, Brown P, Pollo C, Krack P, Nowacki A and Tinkhauser G 2023 Neuromodulation 26 320
[12] Acosta J, Falcone G J, Rajpurkar P and Topol E J 2022 Nat. Med. 28 1773
[13] Feng Y, You H, Zhang Z, Ji R and Gao Y 2019 Proc. AAAI Conf. Artif. Intell. 33 3558
[14] Gao Y, Feng Y, Ji S and Li R 2022 IEEE Trans. Pattern Anal. Mach. Intell. 45 3181
[15] Wang B, Mezlini A M, Demir F, Fiume M, Tu Z, Brudno M, Haibe- Kains B and Goldenberg A 2014 Nat. Methods 11 333
[16] Yang K 2019 Pure Math. 9 1115
[17] Vanden Bulcke C, Wynen M, Detobel J, La Rosa F, Absinta M, Dricot L, Macq B, Cuadra M B and Maggi P 2022 Neuroimage Clin. 36 103252
[18] Estéban O, Markiewicz C, Blair R, Moodie C, Isik A I, Erramuzpe A, Kent J D, Goncalves M, DuPré E, Snyder M, Oya H, Ghosh S, Wright J, Durnez J, Poldrack R A and Gorgolewski K J 2018 bioRxiv
[19] Guimaraes A A R, Abdou T and Bas?ar A 2019 undefined
[20] Estéban O, Gorgolewski K J, Blair R, Berleant S L, Moodie C and Poldrack R A 2017 F1000Res. 6 1498
[21] Yue G, Deng A, Qu Y, Cui H and Wang X 2023 Inf. Sci. 633 182
[22] Wang X J, Gu H G, Jia Y B, et al. 2023 Chin. Phys. B 32 090502
[23] Ni H J, Song Z J, Qin J L, et al. 2025 Chin. Phys. B 34 080504

[1]	Evolutionary hypergraph dismantling via deep reinforcement learning Junjie Qian(钱俊杰), Wenlan Wang(王文蓝), Hanyun Wang(王涵韵), Qiqi Wang(王萁淇), Yao Zhang(张瑶), and Huijia Li(李慧嘉). Chin. Phys. B, 2026, 35(6): 068901.
[2]	A node importance prediction algorithm based on graph attention and contrastive learning Jun Ai(艾均), Yuming Zhang(张玉明), Zhan Su(苏湛), Chenye Guo(郭晨晔), and Mingsong Li(李铭松). Chin. Phys. B, 2026, 35(5): 058901.
[3]	Explosive synchronization and hysteresis in FitzHugh-Nagumo neural networks with higher-order interactions Wen-Xin Cao(曹文鑫), Mao-Sheng Wang(汪茂胜), Fei Xu(徐飞), Shou-Fang Huang(黄守芳), and Ji-Qian Zhang(张季谦). Chin. Phys. B, 2026, 35(3): 030503.
[4]	Six-degree gravity centrality for detecting influential nodes in networks Jianbo Wang(王建波), Bohang Lin(林渤杭), Zhanwei Du(杜占玮), Ping Li(李平), and Xiao-Ke Xu(许小可). Chin. Phys. B, 2025, 34(8): 088902.
[5]	Critical station identification of metro networks based on the integrated topological-functional algorithm: A case study of Chengdu Zi-Qiang Zeng(曾自强), Sheng-Jie He(何圣洁), and Wang Tian(田旺). Chin. Phys. B, 2025, 34(2): 028902.
[6]	Dynamic modeling and analysis of brucellosis on metapopulation network: Heilongjiang as cases Xin Pei(裴鑫), Xuan-Li Wu(武绚丽), Pei Pei(裴沛), Ming-Tao Li(李明涛), Juan Zhang(张娟), and Xiu-Xiu Zhan(詹秀秀). Chin. Phys. B, 2025, 34(1): 018904.
[7]	Dynamic partition of urban network considering congestion evolution based on random walk Zhen-Tong Feng(冯振通), Lele Zhang(张乐乐), Yong-Hong Wu(吴永洪), and Mao-Bin Hu(胡茂彬). Chin. Phys. B, 2025, 34(1): 018902.
[8]	Detecting the core of a network by the centralities of the nodes Peijie Ma(马佩杰), Xuezao Ren(任学藻), Junfang Zhu(朱军芳), and Yanqun Jiang(蒋艳群). Chin. Phys. B, 2024, 33(8): 088903.
[9]	Influencer identification of dynamical networks based on an information entropy dimension reduction method Dong-Li Duan(段东立), Si-Yuan Ji(纪思源), and Zi-Wei Yuan(袁紫薇). Chin. Phys. B, 2024, 33(4): 040502.
[10]	A multilayer network diffusion-based model for reviewer recommendation Yiwei Huang(黄羿炜), Shuqi Xu(徐舒琪), Shimin Cai(蔡世民), and Linyuan Lü(吕琳媛). Chin. Phys. B, 2024, 33(3): 038901.
[11]	Epidemic threshold influenced by non-pharmaceutical interventions in residential university environments Zechao Lu(卢泽超), Shengmei Zhao(赵生妹), Huazhong Shu(束华中), and Long-Yan Gong(巩龙延). Chin. Phys. B, 2024, 33(2): 028707.
[12]	Identification of key recovering node for spatial networks Zijian Yan(严子健), Yongxiang Xia(夏永祥), Lijun Guo(郭丽君), Lingzhe Zhu(祝令哲), Yuanyuan Liang(梁圆圆), and Haicheng Tu(涂海程). Chin. Phys. B, 2023, 32(6): 068901.
[13]	Topological phase transition in network spreading Fuzhong Nian(年福忠) and Xia Zhang(张霞). Chin. Phys. B, 2023, 32(3): 038901.
[14]	Effects of heterogeneous adoption thresholds on contact-limited social contagions Dan-Dan Zhao(赵丹丹), Wang-Xin Peng(彭王鑫), Hao Peng(彭浩), and Wei Wang(王伟). Chin. Phys. B, 2022, 31(6): 068906.
[15]	A novel similarity measure for mining missing links in long-path networks Yijun Ran(冉义军), Tianyu Liu(刘天宇), Tao Jia(贾韬), and Xiao-Ke Xu(许小可). Chin. Phys. B, 2022, 31(6): 068902.

No Suggested Reading articles found!

Viewed

Full text

Abstract

Cited

Metrics
Related Articles

A hypergraph neural network-based framework for identification of Parkinson’s disease subtypes

Cite this article:

Online attention