notesum.ai

Published at November 26

Epidemiology-informed Graph Neural Network for Heterogeneity-aware Epidemic Forecasting

cs.LG

Released Date: November 26, 2024

Authors: Yufan Zheng¹, Wei Jiang², Alexander Zhou³, Nguyen Quoc Viet Hung⁴, Choujun Zhan⁵, Tong Chen²

Aff.: ¹City University of Hong Kong, Hong Kong, China; ²The University of Queensland, Brisbane, Australia; ³The Hong Kong Polytechnic University, Hong Kong, China; ⁴Griffith University, Brisbane, Australia; ⁵South China Normal University, Guangzhou, China

Arxiv: http://arxiv.org/abs/2411.17372v1

	Japan-Prefectures				US-Regions				US-States
RMSE( $\downarrow$ )	2	3	4	5	2	3	4	5	2	3	4	5
AR	1669.898	2071.535	2312.712	2440.119	625.620	813.453	956.083	1078.889	172.179	210.689	239.205	264.437
ARMA	1661.222	2070.781	2312.378	2438.737	622.393	804.126	939.212	1057.998	172.106	207.789	239.828	262.015
GAR	1484.372	1990.001	2281.875	2427.555	607.232	790.684	934.327	1053.783	161.848	199.566	220.901	247.398
VAR	1577.216	2100.213	2443.453	2566.043	694.516	864.829	1000.032	1112.857	297.669	306.723	318.044	333.312
RNN	1198.314	1551.096	1720.115	1851.500	572.438	753.568	890.801	979.708	158.949	195.903	210.525	224.040
LSTM	1255.470	1555.815	1710.241	1839.544	570.277	743.945	913.847	1048.754	161.029	197.831	216.700	226.667
GRU	1181.230	1498.437	1632.849	1781.807	570.898	741.786	893.440	1024.078	159.298	195.385	217.282	247.858
RNN-Attn	1311.157	1751.819	2056.583	2266.540	566.375	757.980	894.176	999.389	162.821	193.795	219.705	243.932
CNNRNN-Res	1401.394	1889.815	2260.501	2570.203	594.910	782.389	873.559	1006.251	228.275	290.916	308.268	320.370
LSTNet	1441.367	2030.374	2331.407	2385.490	606.540	841.441	1030.450	1134.228	195.949	232.584	255.764	276.306
STGCN	1273.563	1389.538	1498.233	1399.753	726.945	814.875	894.509	965.067	205.060	226.799	246.035	260.379
MGNN	1756.653	1922.766	1822.108	1943.240	1472.420	1415.071	1473.778	1487.338	178.034	189.710	198.840	205.517
TMGNN	1500.051	1471.865	1568.222	1558.965	656.364	755.206	829.295	915.758	178.709	188.653	201.361	190.089
ATMGNN	2436.255	2465.356	2473.451	2464.829	1553.142	1607.391	1691.877	1688.661	307.932	374.719	383.963	355.191
Cola-GNN	1167.715	1416.884	1478.782	1572.566	551.810	681.407	793.208	870.818	147.700	175.602	194.336	223.540
Epi-Cola-GNN	1118.387	2056.956	2282.738	1543.628	546.637	747.352	881.656	871.939	148.455	184.681	204.389	219.132
Epi-GNN	1503.244	1296.059	1284.933	1447.863	563.328	708.990	790.972	881.026	149.515	168.939	184.500	190.005
HeatGNN	1148.850	1260.150	1295.697	1378.207	540.573	693.655	781.513	852.311	142.433	164.788	177.624	186.280
PCC( $\uparrow$ )	2	3	4	5	2	3	4	5	2	3	4	5
AR	0.743	0.567	0.413	0.303	0.921	0.868	0.829	0.788	0.934	0.903	0.876	0.850
ARMA	0.745	0.567	0.413	0.303	0.920	0.870	0.827	0.786	0.933	0.902	0.873	0.849
GAR	0.803	0.613	0.444	0.318	0.924	0.870	0.823	0.782	0.938	0.906	0.884	0.861
VAR	0.771	0.564	0.369	0.229	0.895	0.839	0.775	0.715	0.779	0.769	0.749	0.730
RNN	0.887	0.810	0.775	0.757	0.934	0.886	0.848	0.819	0.944	0.915	0.902	0.890
LSTM	0.884	0.831	0.818	0.820	0.933	0.890	0.853	0.820	0.943	0.915	0.897	0.892
GRU	0.894	0.840	0.827	0.807	0.933	0.887	0.849	0.817	0.944	0.916	0.897	0.870
RNN-Attn	0.865	0.735	0.589	0.497	0.936	0.889	0.849	0.820	0.943	0.916	0.891	0.869
CNNRNN-Res	0.828	0.653	0.459	0.239	0.925	0.869	0.833	0.777	0.887	0.802	0.769	0.731
LSTNet	0.820	0.586	0.384	0.338	0.934	0.877	0.795	0.733	0.924	0.885	0.848	0.811
STGCN	0.892	0.874	0.852	0.872	0.899	0.866	0.848	0.831	0.901	0.882	0.861	0.838
MGNN	0.688	0.633	0.644	0.507	0.558	0.557	0.557	0.550	0.926	0.914	0.909	0.901
TMGNN	0.827	0.836	0.833	0.822	0.913	0.885	0.870	0.833	0.925	0.916	0.906	0.913
ATMGNN	0.160	0.156	0.162	0.170	0.496	0.481	0.464	0.456	0.737	0.640	0.633	0.698
Cola-GNN	0.906	0.845	0.870	0.834	0.939	0.903	0.872	0.847	0.953	0.936	0.922	0.904
Epi-Cola-GNN	0.912	0.580	0.444	0.859	0.940	0.889	0.851	0.858	0.953	0.922	0.906	0.909
Epi-GNN	0.821	0.911	0.911	0.869	0.937	0.904	0.885	0.858	0.952	0.937	0.925	0.923
HeatGNN	0.917	0.907	0.910	0.884	0.941	0.908	0.885	0.866	0.953	0.939	0.929	0.921