A GA-BP neural network model based on spectrum-effect relationship for assessing spectrum-effect score and quality evaluation of Cassia seeds extract

doi:10.12122/j.issn.1673-4254.2025.10.05

Abstract

Abstract:

Objective To construct a GA-BP neural network model based on the spectrum-effect relationship of Cassia seeds extract and test its performance for quality control of Cassia seeds using spectrum-effect score. Methods The HPLC fingerprints of Cassia seeds extract (0.1, 0.2, and 0.4 g/mL) were established. In a mouse model of 5-Fu-induced liver injury treated with 0.4, 0.8, and 1.6 g/kg of Cassia seeds extract, the pharmacodynamics parameters were measured to calculate the comprehensive efficacy using AHP-EWM. A GA-BP neural network model between the fingerprints and comprehensive efficacy was constructed, and the corresponding predicted comprehensive efficacy was obtained. The spectrum-effect relationship between the fingerprints and the measured and predicted comprehensive efficacy was established using grey correlation method followed by Gaussian fitting analysis. The spectral efficiency score was calculated using the relative peak area of the fingerprints and the correlation degree of the spectral efficiency. The reliability of the data was tested using the Z-ratio score method. The limit range of the spectral efficiency score was determined and the quality of the verification samples was evaluated. Results The error between the predicted value using the GA-BP neural network model and the measured value of the comprehensive efficacy was less than 0.2. Gaussian fitting analysis showed good fitting between the spectrum-effect relationship data of the measured and predicted comprehensive efficacy. The limit of the spectral efficiency score was 6.16-7.30. The prediction results for each verification group were consistent with the experimental results and within the limit of spectral efficiency score, and the results of Z-ratio score analysis demonstrated good data reliability. Conclusion The GA-BP neural network model can effectively predict the comprehensive efficacy of Cassia seeds extract, and the established spectrum-effect scoring method can be used for quality evaluation of samples.

Key words: GA-BP neural network, Cassia seeds extract, spectrum-effect relationship

Haiyan YAN, Heng WANG, Chuncai ZOU. A GA-BP neural network model based on spectrum-effect relationship for assessing spectrum-effect score and quality evaluation of Cassia seeds extract[J]. Journal of Southern Medical University, 2025, 45(10): 2092-2103.

Figures/Tables 17

References 30

[1]	鄢海燕, 邹纯才. 基于谱效整合分子对接技术研究瓜蒌子及其壳、仁抗凝血的物质基础及作用机制[J].中国中药杂志, 2022, 47(5):1370-82.
[2]	李柯翱, 米尔扎提•麦麦提, 张明惠, 等. 基于统计学分析的寒喘祖帕颗粒HPLC指纹图谱研究[J].中国现代中药, 2022, 24(8): 1555-60.
[3]	李宗民, 朱浩东, 于丹, 等.基于多元统计学分析葛根指纹图谱研究[J].辽宁中医药大学学报,2020,22(11):50-3.
[4]	吴秀平, 徐明珠, 孙宜春, 等. 基于UPLC多指标成分定量和指纹图谱结合化学模式识别法评价不同基原十大功劳木的质量[J]. 中草药,2025, 56(8): 2914-24.
[5]	姜春凤, 孔雪, 周星宇, 等. 基于指纹图谱、多成分含量测定及化学模式识别的柴胡郁金香颗粒质量评价方法研究[J]. 药物分析杂志,2025, 45(2): 201-8.
[6]	董佳乐, 吴晓康, 杨宛颐, 等. 基于神经网络和多元统计分析的动态预测建模分析[J].数字技术与应用, 2024, 42(4): 205-7.
[7]	王靖霖, 王东升, 苑宏英, 等. 指纹图谱技术与人工智能在污染物溯源解析中的应用研究[J]. 环境保护科学, 2022, 48(6): 130-7.
[8]	王恒, 李梦奇, 李燊星, 等. 决明子总蒽醌提取物抗氟尿嘧啶致小鼠肝损伤的谱效关系[J]. 南方医科大学学报, 2023, 43(5): 825-31.
[9]	王恒.决明子对氟尿嘧啶致小鼠肝损伤的保护作用及相关机制研究[D]. 芜湖: 皖南医学院, 2022.
[10]	韩愈琪, 刘雪莱, 上官文斌. 基于本构神经网络橡胶隔振器动态特性建模[J]. 噪声与振动控制, 2023, 43(3): 265-70.
[11]	陈梅妹, 王洋, 雷黄伟,等. 基于多种机器学习算法和语音情绪特征的阈下抑郁辨识模型构建[J]. 南方医科大学学报, 2025, 45(4): 711-7.
[12]	俞炎彧, 邹纯才, 鄢海燕, 等. 基于谱效关联及AHP-EWM 综合赋权的Box-Behnken 响应面法结合GA-BP多指标优化白芍-甘草药对提取工艺[J]. 中国现代应用药学, 2025, 42(3): 410-23.
[13]	汤瑜晨, 胡云飞, 郭抒博, 等. Box-Behnken响应面法结合基准关联度和AHP-EWM优化经典名方清金化痰汤的提取工艺[J]. 分析测试学报, 2024, 43(11): 1794-802, 1812.
[14]	邹纯才, 洪国君, 鄢海燕. 不同数据标准化方法对瓜蒌滴丸抗凝血作用谱效关系的影响[J]. 中国中药杂志, 2018, 43(9): 1864-70.
[15]	史精干, 李燊星, 王恒, 等. 决明子分散片对大鼠急性肝损伤的保护作用及其谱效关系研究[J]. 中国药学杂志, 2023, 58(14): 1295-304.
[16]	黄莉,程索婷,刘竹,等.决明子脐贴对慢传输型便秘大鼠的治疗作用及谱效关系[J]. 南方医科大学学报, 2024, 44(4): 720-6.
[17]	孟宪生, 包永睿, 王帅, 等. 复方中药质量标志物的发现与量效色卡可视化技术[J]. 药学学报, 2019, 54(2): 222-7.
[18]	孟宪生,罗曦,贾梦楠,等.中药质量评价研究现状及“质-量”双标评价方法探讨[J]. 中草药, 2023, 54(22): 7281-6.
[19]	鄢海燕, 魏美玲, 吴雪, 等. 药学本科毕业设计“双导师制”协同培养模式的构建与教学评价[J]. 通化师范学院学报, 2023, 44(8):123-32.
[20]	李雪转. Z比分数法在实验室能力验证中运用的探讨[J]. 计量与测试技术, 2023, 50(7): 86-8.
[21]	鄢海燕, 邹纯才. 《中国药典》（2010年版~2020年版）中药指纹（特征）图谱应用进展与展望[J]. 南方医科大学学报, 2022, 42(1): 150-5.
[22]	Li H, Zhao H, Chen L, et al. Spectrum-effect relationship between HPLC fingerprints and antioxidant activity of Qi-Fu-Yin based on multiple statistical correlation analysis[J]. Phytochem Anal, 2024, 35(7): 1565-76. doi：10.1002/pca.3396
[23]	Liu G, Zhou L, Mu K, et al. Based on spectrum-effect relationship, network pharmacology, and molecular docking, the material basis and mechanism of antioxidant effect of Miao medicine Indigofera stachyoides lindl were studied[J]. ACS Omega, 2024, 9(41): 42199-211. doi：10.1021/acsomega.4c04212
[24]	鄢海燕, 邹纯才. 基于功能主治及其物质基础的中药多元化质量控制模式探讨[J]. 国际药学研究杂志, 2019, 46(4): 266-9.
[25]	舒宇航, 简祈盼, 郑宇锟, 等. 基于灰色关联度与正交偏最小二乘法分析的黄连抗菌谱效关系研究[J].时珍国医国药, 2023, 34(9):2298-302.
[26]	王春艳, 侯益民, 李汉伟, 等. 基于偏最小二乘回归法的黑果菝葜乙酸乙酯部位抗肿瘤谱效关系[J].中成药, 2021, 43(11): 3227-31.
[27]	王志萍,李林杰,王昱涵,等.基于灰色关联度的金母颗粒抗炎作用的谱效关系研究[J].中药新药与临床药理, 2023, 34(5): 675-83.
[28]	张国玉, 熊继东, 魏家艳, 等.灰色关联度分析清半夏体外抗氧化作用的谱效关系[J].华西药学杂志, 2024, 39(4): 382-6.
[29]	裴庆庆, 刘慧慧. 基于高斯函数拟合的多维数据三维可视化仿真[J].计算机仿真, 2024, 41(1): 206-10.
[30]	王彩玲, 张国浩, 闫晶晶. 基于人工神经网络的透射水体亚硝酸盐含量模拟估计[J]. 中国无机分析化学, 2024, 14(7): 857-65.

Evaluation	ALT	AST	MPO	T-AOC	SOD	Weight (g)	Liver index
ALT	1	1	3	3	3	5	5
AST	1	1	3	3	3	5	5
MPO	1/3	1/3	1	1	1	2	2
T-AOC	1/3	1/3	1	1	1	2	2
SOD	1/3	1/3	1	1	1	2	2
Weight (g)	1/5	1/5	1/2	1/2	1/2	1	1
Liver index	1/5	1/5	1/2	1/2	1/2	1	1

Evaluation	ALT	AST	MPO	T-AOC	SOD	Weight (g)	Liver index
ALT	1	1	3	3	3	5	5
AST	1	1	3	3	3	5	5
MPO	1/3	1/3	1	1	1	2	2
T-AOC	1/3	1/3	1	1	1	2	2
SOD	1/3	1/3	1	1	1	2	2
Weight (g)	1/5	1/5	1/2	1/2	1/2	1	1
Liver index	1/5	1/5	1/2	1/2	1/2	1	1

Category	Group	Pharmacodynamic indicators and their standardized values (BZ)							Comprehensive pharmacological effects
Category	Group	Weight (g) /BZ	Liver index (mg·g^-1)/BZ	ALT(U·L^-1) /BZ	AST(U·L^-1) /BZ	MPO ( U·gprot^-1) /BZ	T-AOC(mmol· gprot^-1)/BZ	SOD/U·mgprot^-1 /BZ	ED	MF	MB	MV
Used for modeling	S1-L1	20.38/0.5821	62.49/0.7192	11.49/0.3941	28.81/0.8140	0.475/0.5474	0.074/0.3936	340.2/0.7336	0.5946	0.5803	0.5480
	S2-L2	18.91/0.5401	65.47/0.6864	8.53/0.5309	30.10/0.7790	1.206/0.2156	0.090/0.4787	364.3/0.7856	0.5641	0.5494	0.5419
	S3-L3	20.33/0.5807	63.13/0.7118	13.56/0.3340	38.75/0.6052	0.513/0.5068	0.083/0.4415	398.9/0.8603	0.5054	0.5008	0.4951
	S4-L4	22.54/0.6438	56.33/0.7978	8.10/0.5589	29.77/0.7878	0.525/0.4952	0.066/0.3510	388.6/0.8380	0.6277	0.5009	0.5928
	S5-L5	23.10/0.6598	54.72/0.8212	12.88/0.3516	39.96/0.5869	0.413/0.6295	0.072/0.3830	382.6/0.8251	0.5282	0.5008	0.5107
	S6-L6	21.10/0.6027	55.08/0.8159	13.71/0.3303	46.38/0.5056	0.504/0.5159	0.088/0.4681	403.3/0.8698	0.4766	0.6478	0.4723
	S7-L7	21.11/0.6030	57.69/0.7790	9.26/0.4890	27.59/0.8502	0.888/0.2928	0.081/0.4308	424.9/0.9164	0.5970	0.6208	0.5458
	S8-L8	22.67/0.6475	60.67/0.7407	10.57/0.4286	25.57/0.9170	0.534/0.4869	0.069/0.3670	422.5/0.9111	0.6370	0.5008	0.6126
	S9-W1	20.28/0.5793	56.17/0.8000	8.63/0.5244	35.35/0.6633	0.351/0.7407	0.098/0.5212	296.3/0.6389	0.6227	0.5583	0.5432
	S10-W2	24.04/0.6867	60.49/0.7429	9.24/0.4901	30.83/0.7607	0.366/0.7104	0.092/0.4893	405.2/0.8739	0.6528	0.6334	0.5801
	S11-W3	20.66/0.5901	55.83/0.8049	8.96/0.5056	31.67/0.7405	0.805/0.3230	0.095/0.5053	408.8/0.8815	0.5733	0.5193	0.5257
	S12-W4	22.07/0.6304	58.01/0.7747	12.03/0.3764	38.04/0.6165	0.360/0.7222	0.094/0.5000	415.3/0.8958	0.5710	0.6762	0.5025
	S13-W5	19.14/0.5467	62.94/0.7140	9.81/0.4614	32.59/0.7197	0.463/0.5616	0.097/0.5159	407.6/0.8791	0.5948	0.5884	0.5525
	S14-W6	19.86/0.5673	57.65/0.7795	6.32/0.7167	32.90/0.7128	0.481/0.5405	0.087/0.4627	424.9/0.9164	0.6601	0.5852	0.6391
	S15-W7	21.43/0.6121	54.58/0.8234	19.73/0.2295	32.08/0.7130	0.440/0.5909	0.084/0.4468	405.5/0.8744	0.5391	0.5336	0.5086
	S16-W8	21.63/0.6178	60.53/0.7424	11.24/0.4027	26.82/0.8744	0.369/0.7046	0.089/0.4734	388.4/0.8377	0.6590	0.6456	0.5770
	S17-H1	20.56/0.5873	63.46/0.7081	11.42/0.3966	27.64/0.8484	0.327/0.7951	0.099/0.5266	426.9/0.9207	0.6692	0.6773	0.6855
	S18-H2	19.28/0.5507	55.78/0.8056	6.55/0.6916	34.28/0.6842	0.357/0.7283	0.101/0.5372	347.7/0.7498	0.6775	0.6769	0.5481
	S19-H3	19.25/0.5498	52.71/0.8526	15.01/0.3017	35.42/0.6620	0.413/0.6295	0.112/0.5957	437.0/0.9425	0.5519	0.5149	0.4823
	S20-H4	18.52/0.5290	51.59/0.8711	16.04/0.2822	28.38/0.8264	0.563/0.4618	0.094/0.5000	396.7/0.8556	0.5608	0.6472	0.5288
	S21-H5	20.71/0.5915	64.65/0.6951	8.99/0.5039	30.88/0.7595	0.501/0.5190	0.104/0.5532	444.1/0.9578	0.6208	0.6744	0.6463
	S22-H6	20.55/0.5870	60.00/0.7490	7.32/0.6187	30.95/0.7578	0.428/0.6075	0.098/0.5212	392.0/0.8455	0.6635	0.6693	0.6947
	S23-H7	21.24/0.6067	50.14/0.8963	19.03/0.2379	34.25/0.6848	0.696/0.3736	0.109/0.5797	406.0/0.8755	0.4953	0.6640	0.4724
	S24-H8	21.54/0.6153	60.09/0.7479	6.81/0.6647	24.73/0.9482	0.398/0.6533	0.096/0.5106	412.7/0.8901	0.7533	0.6772	0.7284
Used for verification	Z1-ZL1	19.17/0.5476	66.50/0.6758	22.26/0.2034	33.31/0.7041	1.162/0.2238	0.076/0.4042	393.8/0.8492	0.4437			0.5664
	Z2-ZL2	21.54/0.6153	63.41/0.7087	12.04/0.3762	32.40/0.7237	0.584/0.4452	0.071/0.3776	332.6/0.7174	0.5345			0.4729
	Z3-ZW1	20.93/0.5978	59.52/0.7550	15.33/0.2953	27.40/0.8558	0.608/0.4276	0.079/0.4202	414.9/0.8949	0.5607			0.4919
	Z4-ZW2	22.38/0.6392	59.90/0.7502	15.39/0.2943	30.12/0.7786	0.375/0.6933	0.088/0.4681	409.9/0.8839	0.5788			0.5526
	Z5-ZH1	21.11/0.6030	63.81/0.7043	8.046/0.5628	38.50/0.6091	0.496/0.5242	0.095/0.5053	407.7/0.8792	0.5885			0.4762
	Z6-ZH2	19.72/0.5633	52.36/0.8583	6.509/0.6956	28.24/0.8304	0.260/1.000	0.089/0.4734	381.0/0.8216	0.7710			0.6579

Category	Group	Pharmacodynamic indicators and their standardized values (BZ)							Comprehensive pharmacological effects
Category	Group	Weight (g) /BZ	Liver index (mg·g^-1)/BZ	ALT(U·L^-1) /BZ	AST(U·L^-1) /BZ	MPO ( U·gprot^-1) /BZ	T-AOC(mmol· gprot^-1)/BZ	SOD/U·mgprot^-1 /BZ	ED	MF	MB	MV
Used for modeling	S1-L1	20.38/0.5821	62.49/0.7192	11.49/0.3941	28.81/0.8140	0.475/0.5474	0.074/0.3936	340.2/0.7336	0.5946	0.5803	0.5480
	S2-L2	18.91/0.5401	65.47/0.6864	8.53/0.5309	30.10/0.7790	1.206/0.2156	0.090/0.4787	364.3/0.7856	0.5641	0.5494	0.5419
	S3-L3	20.33/0.5807	63.13/0.7118	13.56/0.3340	38.75/0.6052	0.513/0.5068	0.083/0.4415	398.9/0.8603	0.5054	0.5008	0.4951
	S4-L4	22.54/0.6438	56.33/0.7978	8.10/0.5589	29.77/0.7878	0.525/0.4952	0.066/0.3510	388.6/0.8380	0.6277	0.5009	0.5928
	S5-L5	23.10/0.6598	54.72/0.8212	12.88/0.3516	39.96/0.5869	0.413/0.6295	0.072/0.3830	382.6/0.8251	0.5282	0.5008	0.5107
	S6-L6	21.10/0.6027	55.08/0.8159	13.71/0.3303	46.38/0.5056	0.504/0.5159	0.088/0.4681	403.3/0.8698	0.4766	0.6478	0.4723
	S7-L7	21.11/0.6030	57.69/0.7790	9.26/0.4890	27.59/0.8502	0.888/0.2928	0.081/0.4308	424.9/0.9164	0.5970	0.6208	0.5458
	S8-L8	22.67/0.6475	60.67/0.7407	10.57/0.4286	25.57/0.9170	0.534/0.4869	0.069/0.3670	422.5/0.9111	0.6370	0.5008	0.6126
	S9-W1	20.28/0.5793	56.17/0.8000	8.63/0.5244	35.35/0.6633	0.351/0.7407	0.098/0.5212	296.3/0.6389	0.6227	0.5583	0.5432
	S10-W2	24.04/0.6867	60.49/0.7429	9.24/0.4901	30.83/0.7607	0.366/0.7104	0.092/0.4893	405.2/0.8739	0.6528	0.6334	0.5801
	S11-W3	20.66/0.5901	55.83/0.8049	8.96/0.5056	31.67/0.7405	0.805/0.3230	0.095/0.5053	408.8/0.8815	0.5733	0.5193	0.5257
	S12-W4	22.07/0.6304	58.01/0.7747	12.03/0.3764	38.04/0.6165	0.360/0.7222	0.094/0.5000	415.3/0.8958	0.5710	0.6762	0.5025
	S13-W5	19.14/0.5467	62.94/0.7140	9.81/0.4614	32.59/0.7197	0.463/0.5616	0.097/0.5159	407.6/0.8791	0.5948	0.5884	0.5525
	S14-W6	19.86/0.5673	57.65/0.7795	6.32/0.7167	32.90/0.7128	0.481/0.5405	0.087/0.4627	424.9/0.9164	0.6601	0.5852	0.6391
	S15-W7	21.43/0.6121	54.58/0.8234	19.73/0.2295	32.08/0.7130	0.440/0.5909	0.084/0.4468	405.5/0.8744	0.5391	0.5336	0.5086
	S16-W8	21.63/0.6178	60.53/0.7424	11.24/0.4027	26.82/0.8744	0.369/0.7046	0.089/0.4734	388.4/0.8377	0.6590	0.6456	0.5770
	S17-H1	20.56/0.5873	63.46/0.7081	11.42/0.3966	27.64/0.8484	0.327/0.7951	0.099/0.5266	426.9/0.9207	0.6692	0.6773	0.6855
	S18-H2	19.28/0.5507	55.78/0.8056	6.55/0.6916	34.28/0.6842	0.357/0.7283	0.101/0.5372	347.7/0.7498	0.6775	0.6769	0.5481
	S19-H3	19.25/0.5498	52.71/0.8526	15.01/0.3017	35.42/0.6620	0.413/0.6295	0.112/0.5957	437.0/0.9425	0.5519	0.5149	0.4823
	S20-H4	18.52/0.5290	51.59/0.8711	16.04/0.2822	28.38/0.8264	0.563/0.4618	0.094/0.5000	396.7/0.8556	0.5608	0.6472	0.5288
	S21-H5	20.71/0.5915	64.65/0.6951	8.99/0.5039	30.88/0.7595	0.501/0.5190	0.104/0.5532	444.1/0.9578	0.6208	0.6744	0.6463
	S22-H6	20.55/0.5870	60.00/0.7490	7.32/0.6187	30.95/0.7578	0.428/0.6075	0.098/0.5212	392.0/0.8455	0.6635	0.6693	0.6947
	S23-H7	21.24/0.6067	50.14/0.8963	19.03/0.2379	34.25/0.6848	0.696/0.3736	0.109/0.5797	406.0/0.8755	0.4953	0.6640	0.4724
	S24-H8	21.54/0.6153	60.09/0.7479	6.81/0.6647	24.73/0.9482	0.398/0.6533	0.096/0.5106	412.7/0.8901	0.7533	0.6772	0.7284
Used for verification	Z1-ZL1	19.17/0.5476	66.50/0.6758	22.26/0.2034	33.31/0.7041	1.162/0.2238	0.076/0.4042	393.8/0.8492	0.4437			0.5664
	Z2-ZL2	21.54/0.6153	63.41/0.7087	12.04/0.3762	32.40/0.7237	0.584/0.4452	0.071/0.3776	332.6/0.7174	0.5345			0.4729
	Z3-ZW1	20.93/0.5978	59.52/0.7550	15.33/0.2953	27.40/0.8558	0.608/0.4276	0.079/0.4202	414.9/0.8949	0.5607			0.4919
	Z4-ZW2	22.38/0.6392	59.90/0.7502	15.39/0.2943	30.12/0.7786	0.375/0.6933	0.088/0.4681	409.9/0.8839	0.5788			0.5526
	Z5-ZH1	21.11/0.6030	63.81/0.7043	8.046/0.5628	38.50/0.6091	0.496/0.5242	0.095/0.5053	407.7/0.8792	0.5885			0.4762
	Z6-ZH2	19.72/0.5633	52.36/0.8583	6.509/0.6956	28.24/0.8304	0.260/1.000	0.089/0.4734	381.0/0.8216	0.7710			0.6579

Group	General model gauss	Coefficients (with 95% confidence bounds)			Goodness of fit
Group	General model gauss	a	b	c	SSE	R-square	Adjusted R-square	RMSE
Ⅰ	f(x) = a*exp(-((x-b)/c)²)	0.7850	1.1150	0.7935	0.0015	0.9927	0.9921	0.0073
Ⅱ		0.8244	1.0330	0.7339	0.0006	0.9977	0.9975	0.0047
Ⅲ		1.3500	1.4890	0.9330	0.0015	0.9961	0.9959	0.0074