Skip to main content

Decoding the underlying mechanisms of Di-Tan-Decoction in treating intracerebral hemorrhage based on network pharmacology



Chinese medicine usually acts as "multi-ingredients, multi-targets and multi-pathways" on complex diseases, and these action modes reflect the coordination and integrity of the treatment process with traditional Chinese medicine (TCM). System pharmacology is developed based on the cross-disciplines of directional pharmacology, system biology, and mathematics, has the characteristics of integrity and synergy in the treatment process of TCM. Therefore, it is suitable for analyzing the key ingredients and mechanisms of TCM in treating complex diseases. Intracerebral Hemorrhage (ICH) is one of the leading causes of death in China, with the characteristics of high mortality and disability rate. Bring a significant burden on people and society. An increasing number of studies have shown that Chinese medicine prescriptions have good advantages in the treatment of ICH, and Ditan Decoction (DTT) is one of the commonly used prescriptions in the treatment of ICH. Modern pharmacological studies have shown that DTT may play a therapeutic role in treating ICH by inhibiting brain inflammation, abnormal oxidative stress reaction and reducing neurological damage, but the specific key ingredients and mechanism are still unclear.


To solve this problem, we established PPI network based on the latest pathogenic gene data of ICH, and CT network based on ingredient and target data of DTT. Subsequently, we established optimization space based on PPI network and CT network, and constructed a new model for node importance calculation, and proposed a calculation method for PES score, thus calculating the functional core ingredients group (FCIG). These core functional groups may represent DTT therapy for ICH.


Based on the strategy, 44 ingredients were predicted as FCIG, results showed that 80.44% of the FCIG targets enriched pathways were coincided with the enriched pathways of pathogenic genes. Both the literature and molecular docking results confirm the therapeutic effect of FCIG on ICH via targeting MAPK signaling pathway and PI3K-Akt signaling pathway.


The FCIG obtained by our network pharmacology method can represent the effect of DTT in treating ICH. These results confirmed that our strategy of active ingredient group optimization and the mechanism inference could provide methodological reference for optimization and secondary development of TCM.

Peer Review reports


Hypertensive intracerebral hemorrhage (ICH) is one of the main causes of death and disability among Chinese residents. According to the statistics, the mortality rate is as high as 35%-52%, and about 80% of the surviving patients still have disabilities within half a year [1], which brings great burden to individuals and society. ICH mostly occurs in the area of perforator arteries. Under the influence of long-term hypertension, these perforating arteries will have a series of pathological changes, such as "lipid hyaline degeneration", hyperplasia of subendothelial fibroblasts, deposition of macrophages, and replacement of collagen-rich medial smooth muscle cells. These changes result in decreased vascular compliance and lumen stenosis, which is prone to cerebral hemorrhage when blood pressure fluctuates greatly. Modern medicine mainly treats ICH by controlling blood pressure, clearing hematoma and preventing rebleeding. Some basic studies in recent years have revealed some key mechanisms for treating cerebral hemorrhage. Isoliquiritigenin reduces early brain injury following experimental intracerebral hemorrhage by suppressing ROS- and/or NF-B-mediated NLRP3 inflammasome activation via the Nrf2 antioxidant pathway [2]. Following ICH, the Striatal P2X7R and NLRP3 inflammasomes were activated. P2X7R gene silencing inhibited NLRP3 inflammasome activation and interleukin (IL)-1/IL-18 release, significantly alleviating brain edema and neurological deficits [3].

In recent years, many experimental studies have been carried out on ICH. Statins can improve neurological outcome and promote neurovascular recovery after ICH [4]. Mesenchymal stromal cells-derived exosomes effectively improve functional recovery after ICH [5]. However, due to the limited treatments and effects, there are still great challenges to the problems such as the disappearance of brain edema, rebleeding after treatment, and recovery after nerve function injury. Increasing evidence confirms that the prescriptions of TCM have been widely used in the treatment of ICH, Ling et al. investigate that Tongfu Xingshen Decoction can improve clinical symptoms of patients and promote the recovery of neurological function by reducing serum S100β, IL-6 and MMP-9 levels [6]. Guo et al. reported that Angong Niuhuang Pill can alleviate the brain damage caused by ICH from diuresis and dehydration, reducing intracranial pressure and protecting central nervous system [7]. Zhou et al. confirmed that DTT can promote the recovery of cognitive function and improve the clinical efficacy in patients with ICH [8]. Among these prescriptions, DTT is one of the most high frequency used in clinic.

The DTT prescription contains 9 herbs: Glycyrrhiza uralensis Fisch. ex DC.(1.5 g), Zingiber officinale Roscoe (1.5 g), Citrus reticulata Blanco (2.1 g), Arisaema heterophyllum Blume (2.5 g). Panax ginseng C.A.Mey.(3 g), Pinellia ternata (Thunb.) Makino(3 g), Poria cocos (Schw.) Wolf.(3 g), Citrusacida Pers (2.5 g), Bambusa tuldoides Munro (2.1 g).

The prescriptions of TCM have multiple ingredients, multiple targets and multiple pathways, which are necessary factors for treating many complex diseases. As a new subject, network pharmacology is beneficial to identify the effective ingredients and explore the mechanisms of TCM in modern pharmacology research. It is helpful to explain the "combination-effect relationship" and compatibility rule of TCM at a system level. The effect of TCM on cells and organisms is a complex biological network, which is inline with the "integrity" characteristics of systems pharmacology. Thus, system pharmacology is suitable for studying the underlying rules of TCM prescription.

At present, several formula optimization methods have been proposed. Most of these methods take the structural similarity of ingredients as the most important feature but ignore the linkage of pathogenic genes and drug targets. In this study, systems pharmacology was used to detect FCIG and clarify the mechanisms of DTT in treating ICH. We present a reverse optimization model based on the association between disease genes and ingredient targets is proposed, which provides space for optimization based on effective proteins. This method can well determine the optimization space of the target. Second, reverse searching related ingredients based on the optimized space provided by effective protein. The results showed that the enrichment functional pathway of effective proteins could cover 96% of the enrichment functional pathway of disease genes.

Materials and methods

Construct weighted gene regulatory network of ICH

PPI data derived from Dip [9], HPRD [10], BioGRID [11], STRING [12], Reactome [13], Mint [14] and Intact [15] were used to construct comprehensive protein–protein interaction network (Supplementary Table 1). ICH related genes were extracted from GeneCards. The genes with “Relevance Scores” higher than average score were kept as high pathogenic gene. These pathogenic genes are aligned into the PPI network to figure out the weighted gene regulatory network of ICH. The Cytoscape software was utilized to visualize the network.

Collect chemical ingredients of DTT

Ingredients of DTT were extracted from three herbal medicine databases: TCMID and TCMSP. For all ingredients, the OpenBabel toolkit (version 2.4.1) was employed to convert the format of structure to canonical SMILES. Subsequently, the chemical properties, such as molecular weight (MW), DL (drug-likeness), Caco-2 permeability (Caco-2) and oral bioavailability (OB) were retrieved from TCMSP.

Select potential active ingredients of DTT based on ICH models

Three published ADME-related modules, OB, Caco-2 and DL, were used to select the bio-active ingredients. OB (%F) refers to the relative amount and rate of absorption of drugs into circulation after oral administration. High oral bio-availability is one of the key indicators to determine the therapeutic properties of drugs. Selecting suitable ingredients with OB ≥ 30% as potential active ingredients for next step analysis. Caco-2 cell model is a human clonal colon adenocarcinoma cell, which has similar structure and function as intestinal epithelial cells, and it contains an enzyme system related to the brush border epithelium of the small intestine, which can be used to simulate the intestinal transport of drugs in vivo. The transport rate (nm/s) of ingredients in Caco-2 cells represents the permeability of intestinal epithelium. Ingredients with Caco-2 value less than -0.4 have poor permeability in intestinal epithelial cells, so we choose ingredients with Caco-2 > -0.4 as candidate ingredients. DL is used to evaluate the drug-like properties of expected compounds, which is helpful to optimize pharmacokinetics and drug characteristics, such as solubility and chemical stability. In this study, the drug-like scores of the ingredients were more than 0.18 as the selection criteria [16].

After screening by ADME, some ingredients that do not meet the three screening criteria are also selected, because experiments have proved that they have high content and high biological activity. These high-content and high-activity ingredients were merged with ADMET screened ingredients for subsequent analysis.

Predicting targets of active ingredients

To obtain the targets of active ingredients in DTT, the widely used prediction tools, HitPick, Swiss Target Prediction and Similarity Ensemble Approach (SEA) [8] and Swiss Target Prediction were used to predict the targets. All chemical structures were converted to SMILE format.

Node importance calculation method

For calculating the importance of each node in the network, we constructed a Node importance calculation method (BCD), in which, m represent the number of nodes in the network; h, i and j represent nodes in the network, V represent the unit of all nodes in the network, \({\upsigma }_{hi}\) represents the number of the shortest path between nodes h and i; \({\upsigma }_{hi \left(j\right)}\) is the number of the shortest path passing through node j.\({\mathrm{T}}_{h}\) is the number of edges directly connected to a node h. N stands for natural number. The method can be described as follows:

$${\mathrm{BD}}_{h} = {\sqrt{{\mathrm{ T}}_{h} \times ( \sum_{h \ne i \ne j \in V}\frac{{\upsigma }_{hi(j)}}{{\upsigma }_{hi}} )}}$$

There are m nodes in the network. Combining degree with the center degree of betweenness, BD can effectively reflect the direct relationship between nodes and neighboring nodes, as well as the control function of neighboring nodes in the whole network [17,18,19].

After being quantized, BD was sorted from small to large, and was represented by a new variable Y.

$$\mathrm{Y}=\left[{\mathrm{Y}}_{1}, {\mathrm{Y}}_{2}, {\mathrm{Y}}_{3}, \dots , {\mathrm{Y}}_{\mathrm{m}}\right]= \left[{\mathrm{BD}}_{\mathrm{x}}, {\mathrm{BD}}_{\mathrm{x}+1}, {\mathrm{BD}}_{\mathrm{x}+2}, \dots , {\mathrm{BD}}_{\mathrm{x}+\mathrm{n}}\right],\mathrm{ x}\in \left[1,\mathrm{ m}\right]\mathrm{ and x}+\mathrm{n}=\mathrm{m}$$

The new variable R represented the nodes in the network. Each R responds to its unique Y.

$$\mathrm{R }\in \left\{{\mathrm{R}}_{1}, {\mathrm{R}}_{2}, {\mathrm{R}}_{3}, \dots , {\mathrm{R}}_{\mathrm{m}}\right\} \rightleftharpoons \left[{\mathrm{Y}}_{1}, {\mathrm{Y}}_{2}, {\mathrm{Y}}_{3}, \dots , {\mathrm{Y}}_{\mathrm{m}}\right]$$

BCD represented the important nodes selected form all nodes in the network. N represented natural number.

$$\mathrm{BCD}\in \left\{\begin{array}{c}\left\{{\mathrm{R}}_{\mathrm{r}} , {\dots ,{\mathrm{R}}_{(\mathrm{m}-2)} , {\mathrm{R}}_{(\mathrm{m}-1)} ,\mathrm{ R}}_{\mathrm{m}}\right\} \rightleftharpoons \left[{\mathrm{Y}}_{\left(\mathrm{m}+1\right)/2} , {\dots ,{\mathrm{Y}}_{(\mathrm{m}-2)} , {\mathrm{Y}}_{(\mathrm{m}-1)} ,\mathrm{ Y}}_{\mathrm{m}}\right], m=2p+1 and p\in N\\ \\ \\ \left\{{\mathrm{R}}_{\mathrm{r}} ,\dots , {\mathrm{R}}_{(\mathrm{m}-2)} , {\mathrm{R}}_{(\mathrm{m}-1)} , {\mathrm{R}}_{\mathrm{m}}\right\} \rightleftharpoons \left[\frac{({\mathrm{Y}}_{\mathrm{m}/2}+ {\mathrm{Y}}_{(\mathrm{m}+2)/2})}{2} ,\dots , {\mathrm{Y}}_{(\mathrm{m}-2)} , {\mathrm{Y}}_{(\mathrm{m}-1)} , {\mathrm{Y}}_{\mathrm{m}}\right], m=2p and p\in N\end{array}\right.$$

We collected all ingredients of DTT from databases and literature. All ingredients of DTT were used to screen out potential active ingredients by using previous reported ADME models. Three public online tools were employed to identify the targets of the active ingredients. The active ingredients and their targets were used to construct CT network. The pathogenic genes were mapped to a comprehensive protein–protein interaction network to obtain a pathogenic gene regulation network. CT network and pathogenic gene regulation network were integrated and then used to find potential effect space (PES) by using novel node importance calculation model. Finally, the contribution index (CI) model was employed to optimize the FCIG in the PES, and the key functional ingredients were obtained, and the molecular mechanism of DTT on ICH was expounded.

Gene ontology and pathway analysis

Function enrichment analysis is the basic way to understand gene function. Here, the clusterProfiler of R package were performed to conduct pathway and GO term enrichment analysis. The p value less than 0.05 were considered to be functionally related pathways and GO terms.

Experimental validation


Jiangsu Yongjian Biotech Co., Ltd provided 6-shogaol, 6-singerol, kaempferol, and sitostrol (98% purity by HPLC) (Chengdu, China). Gibco supplied fetal bovine serum (FBS) and Dulbecco's modified Eagle's medium (DMEM) (Grand Island, USA). CHI SCIENTIFIC provided the mouse hippocampal HT-22 cells (Shanghai, China). Mitsubishi Gas Chemical Company, Inc. provided the hypoxic bags (Japanese). Dojindo Laboratories supplied the Cell Counting Kit-8 (CCK-8) (Japanese).

Cell Culture and oxygen–glucose deprivation (ogd) treatment

The cells were cultured in DMEM with 10% FBS, and incubated at 37 °C under 5% CO2. Hypoxic bags were used to perform OGD model according to the methods in the literature [20].

Cell viability assay

HT22 cells (6 × 104 cells/well) were seeded in 96-well plates, and treated with different concentrations of 6-Shogaol, 6-Singerol, kaempferol, and sitostrol (40, 80, 120, 160 and 200 μM). CCK8 was superinduced to 96-well plate for 4 h. The plate reader was utilized to detect the absorbance at 450 nm.

Statistical analysis

The data were all expressed as mean SEM. For multiple comparisons, one-way ANOVA was used, and the Student's t test was used to compare the significance of differences between two groups. If the p-value was less than 0.05, the results were considered statistically significant.


In this study, we designed a new system pharmacology module, which was used to detect the key active ingredients and clarify the therapeutic mechanism of DTT in treating ICH. The work flow is illustrated in (Fig. 1) and described as follows: Firstly, all the effective ingredients of DTT were collected from databases and literature. The potential active ingredients were selected from DTT and predicted by three published prediction tools. Then, the weighted gene regulatory network and the active ingredient target network are used to construct the PES to determine the effective protein. Selecting key active ingredients based on CI module by using effective proteins. Finally, the molecular mechanism of DTT in treating ICH was deduced from the FCIG.

Fig. 1
figure 1

The work scheme of system pharmacology approach

Chemical analysis

Chemical identification is the critical step to clarify the material basis and action mechanism of compound prescription. In this study, the information of specific chemical ingredients of DTT Chinese herbal medicines was obtained from the literature (Table 1). The results showed that the chemical ingredients of herbs and the content of identified ingredients provided experimental auxiliary chemical space for searching active ingredients. The chemical constituent analysis serves as a reference for the screening of active constituents in DTT.

Table 1 The information on chemical analysis of the herbs from the literature in DTT

Select potential active ingredients

Although each TCM compound contains many ingredients, only a few ingredients have satisfactory pharmacodynamic and pharmacokinetic properties. In the present work, three ADME-related models, including OB, Caco-2 and DL, are used to screen active ingredients. After ADME screening, some ingredients not pass the criterion of three ADME-related models were kept as active ingredients due to their high content and high bio-activity. Finally, 181 active ingredients were captured from 939 active ingredients (Table 2).

Table 2 Components in DTT for further analysis after AD ME screening

Construct the weighted gene regulatory network of ICH

Constructing and analyzing weighted gene regulatory network is the basis and key step to understand the pathogenesis of ICH and provide intervention strategies. We obtained the genes related to ICH from Genecard database. We selected 2648 genes larger than the average score as the genes related to ICH. At the same time, we used dispenet for comparison, and found that the selected genes could cover 88% of the dispenet library. The comprehensive PPI network, which combined by from several PPI databases were download from the CMGRN [30] and PTHGRN [31]. A total of 999 genes were extracted from DisGeNET and OMIM related to ICH, and mapped to PPI network, and a weighted gene regulatory ne48.82twork was constructed. The weighted gene regulatory network contains 999 nodes and 46,432 edges (Fig. 2). In the disease network, the degree indicates the importance of nodes. We further analyze it by using Network Analyzer software. In PPI network, the target average degree of different ingredients is 48.82. Among the top weighted targets, APP [32], NOTCH3 [33], KRIT1, CCM2, PDCD10, CST3, ENG, TREX1, ATRIP, APOE, etc.

Fig. 2
figure 2

The disease weight gene regulatory network of ICH

Common ingredients of Chinese herbal medicines in DTT

It can be seen from Table 3 that there are 12 active ingredients shared by more than two kinds of herbs in DTT. For example, β-sitosterol is a common ingredient of ginseng, Arisaema cum, orange, ginger and Pinellia ternate. Studies have found that β-sitosterol has a strong inhibitory effect on oxygen free radicals and a strong antioxidant effect on oils and fats [34]. Naringin is shared by Fructus Aurantii Immaturus, Glycyrrhrizae Radix and Exocarpium Citri Grandis. Naringin can obviously alleviate permanent nerve injury and has protective effect. Its mechanism is to down-regulate the expression of NOD2, RIP2 and MMP-9, up-regulate the expression of claudin-6 and protect the blood–brain barrier [35]. 6-gingerol is a common ingredient of Pinellia ternata and Ginger. 6- gingerol may protect PC12 cells against apoptosis induced by Aβ 1–42 through PI3K/Akt/GSK-3β signaling pathway, and has protective effect on nerve cells [36]. In addition, linolenic acid is shared by Pinellia ternata and Arisaema cum laude. Experiments show that increasing DHA level in brain can limit the expression of inflammatory factors after TBI and accelerate the recovery of nerve function [37].

Table 3 12 active ingredients shared by more than two kinds of herbs in DTT

Special ingredients of Chinese herbal medicine in DTT

Apart from the common ingredients, most herbs have their specific ingredients. For example, the main ingredient of ginseng is ginsenosides. Studies by Shi et al. and others have found that neural stem cells are activated and proliferated after intracerebral hemorrhage [38]. Ginsenosides can induce the proliferation of neural stem cells and improve motor function after intracerebral hemorrhage. Ferulic acid is one of the main ingredients of Arisaema cum laude. Ferulic acid reduces the expression of phosphorylated IKK and the transport of NRF2 and NF-kappab to the nucleus, thus inhibiting the activities of IL-6 and NF-kappab promoters. These data indicate that ferulic acid could inhibit inflammatory by mediating IKK/NF-kappab signaling pathway [39]. Naringin (DTT187) is one of the most effective ingredients in tangerine peel. Liu Wei et al. found that naringin pre-intervention can effectively alleviate cerebral ischemia–reperfusion injury, and the protective effect is closely related to the decrease of Cx43 expression in astrocytes [40].

Target prediction of active ingredients

In order to deduce the underlying mechanism of DTT in the treatment of ICH, 181 active ingredients and 1544 target points were employed to construct the ingredient target network. Among them, several active ingredients are associated with multiple targets, resulting in 41,650 target relationships between 181 active ingredients and 1544 targets. The average target number for each ingredient is 47.11. The results indicate that DTT is a multi-ingredient and multi-target therapy for ICH. Among these ingredients, the target of MOL000131 (degree = 333) is the most, followed by MOL000358 (degree = 300), MOL002495 (degree = 246), MOL013156 (degree = 217), MOL000442 (degree = 156) and MOL000432 (degree = 146). Most of these ingredients are related to inflammation and oxidation related pathways of DTT. For example, 6-gingerol has anti-inflammatory and antioxidant effects, which can improve the activities of SOD), glutathione peroxidase and catalase in organisms, eliminate hydroxyl radicals and superoxide radicals, and reduce lipid peroxides in tissues [41]. Stigmasterol can significantly inhibit the increase of ROS induced by Ang ii in A7r5 cells, and increase the activities of SOD and CAT enzymes in A7r5 cells treated by Ang ii. In addition, stigmasterol can obviously inhibit the increased iNOS mRNA and protein levels of COX-2 and iNOS induced by LPS, and has anti-inflammatory effect [42]. The role of other ingredients in the treatment of ICH has been described in the chapters of "Shared Ingredients of Chinese Herbal Medicine in DTT" and "Specific Ingredients of Chinese Herbal Medicine in DTT". These results proved the important role of these ingredients in ICH and further suggested that the multi-ingredient role of DTT in treating ICH.

In the ingredient target network, the average target degree of different targets is 3.86. Among the 20 weighted targets, ABCG2, ABCB1, ALOX5, ALOX15, ALOX12, etc. Interestingly, most of these targets are related to oxidation and inflammation, which have been confirmed to be related to the pathogenesis of ICH, and may indicate the potential therapeutic mechanism of DTT on ICH. For example, caspase-3 is closely related to cerebral hemorrhage and is one of the main factors of neuronal apoptosis. The activation of caspase-3 after intracerebral hemorrhage may be the mechanism of ischemia–reperfusion injury caused by secondary cerebral ischemia around hematoma [43]. In addition, the early release of cytochrome C may be related to hematoma occupation, ischemia and hypoxia of peripheral neurons caused by brain edema, energy metabolism disorder, calcium overload, oxidative stress and the production of a large number of free radicals.The mechanism by which DTT can block the release of cytochrome C is related to the mitochondrial permeability transition pore channel and the Bcl-2 family of proteins. By blocking the release of cytochrome C into the cytoplasm, the apoptosis of neurons was prevented [44].

These results suggest that DTT can treat ICH synergistically by regulating inflammation and antioxidant function, which further confirms the multi-target role of DTT in ICH treatment.

Functional proteins selection and validation based on dtt potential effect space

We constructed a C-T-P network based on PPI using the weighted pathogenic gene regulatory network of ICH and the target network of active ingredients. In this network, it contains 2786 nodes and 45,647 edges. We extracted the relationship between drug targets and pathogenic genes from C-T-P network and constructed DTT potential effect space. Node importance is an important factor to be considered in optimizing network. At present, the methods to describe node importance mainly include degree, betweenness, closeness to centrality, shortest path and so on. These methods mainly depend on the nature of a certain aspect of the network to describe the importance of nodes in the network. In this study, we designed a new network importance calculation method, which takes into account the influence and connectivity of nodes. The comparison results show that the therapeutic response protein obtained by our node importance calculation method accounts for 92.57% of the pathogenic gene enrichment pathway in go function enrichment analysis, which shows that our method has good accuracy. There are three types of proteins in DTT potential effect space. The first is the direct interaction between pathogenic genes and drug targets. We define this category as the basic common targets. The second category is the interaction of disease-specific targets. The third category is the interaction of ingredient-specific targets. In order to determine whether the therapeutic response protein we selected from DTT PES is optimal, we examined the coverage of the enrichment pathway of therapeutic response protein, common target, ingredient-specific target and disease-specific target in the enrichment pathway of pathogenic genes from the functional level. The results showed that the coverage rate of therapeutic response protein was as high as 84.9%, which was 3.1%, 80.6% and 16.8% higher than the three, indicating that the therapeutic response protein we selected had better functional coverage (Fig. 3A and B).

Fig. 3
figure 3

Functional proteins validation. KEGG pathways distribution of response proteins, common targets, ingredient-specific targets, disease-specific targets and pathogenic genes (A); the coverage rate of response proteins, common targets, ingredient-specific targets, and disease-specific targets enriched pathways compared with pathogenic gene-enriched pathways of ICH (B)

The contribution coefficient model is constructed to optimize the effective ingredients and obtain FCIG values, which can be used to clarify the potential mechanism of DTT in treating ICH. According to the result of contribution accumulation rate (Fig. 4), the cumulative contribution rate of the first 44 ingredients reaches 90.75%, which is selected as FCIG. Include EIC (MOL000131), 6-shogaol (MOL002495), [(2r)-2-[[(2r)-2-(benzoylamino)-3-phenylpropyl] amino] methyl]-3-phenylpropyl] acetate (n = 1) Beta-sitosterol (MOL000358), kaempferol (MOL000422), methyl linoleate (MOL001641), etc.

Fig. 4
figure 4

The contribution accumulation rate of DTT FCIG

There is a high consistency between FCIG and C-T network in the number of pathogenic genes. In order to evaluate whether the genes in FCIG are related to ICH, we compared the coverage of pathogenic genes in FCIG and C-T networks. We collected the published literature and the known ICH pathogenic genes in the database, which were verified by the average score of GeneCards greater than 6. It was found that there were 389 pathogenic genes in C-T network and 348 pathogenic genes in FCIG of DTT (Fig. 5A and B). The FCIG of DTT reached 89.5% of C-T network (Fig. 5C). This result shows that our FCIG detection model can maximize the coincidence degree of pathogenic genes in the formula CTT network.

Fig. 5
figure 5

Venn diagram was used to visualize the overlap number of pathogenic genes between C-T network targets (A) and FCIG network targets (B), and the overlap number of C-T network targets overlap pathogenic genes and FCIG network targets overlap pathogenic genes

FCIG is highly consistent with C-T network in gene function level: another index to evaluate the importance of FCIG network is determined by their functional consistency, which can be evaluated by their enrichment pathway in KEGG. The purpose is to test whether FCIG found in DTT can represent its complete C-T network. This result shows that our FCIG detection model can cover C-T network functions to the maximum extent (Fig. 6). There is a high degree of coincidence between FCIG and C-T network in topology: node degree is a key topological parameter to characterize network topology, and it is also the most influential node in the network, so we use it to further determine the importance of FCIG. A mathematical model was established to evaluate the importance of network FCIG in DTT. Then the CC value of FCIG in DTT was calculated by topological parameters. The results show that FCIG can cover 80.44% of C-T composed of DTT. The results show that the FCIG in DTT is consistent with the pathway and pathogenic genes, and our FCIG model can maximize the coverage of network topology and C-T network composed of DTT.

Fig. 6
figure 6

The functional similarity analysis between targets of C-T network and FCIG

Functional core ingredients selection and validation

GO enrichment analysis was performed using the R software clusterProfiler package to identify the biological functions of the main targets with p values < 0.05. To further profile the combined effects of DTT, all targets that interacted with FCIG in DTT were enriched by GO enrichment analysis (Fig. 7).

Fig. 7
figure 7

Go enrichment analysis of the targets of FCIG

DTT regulatory targets were abundantly expressed in biological processes related to inflammatory response, according to GO analysis.For example, reducing oxidative stress (GO:0,070,482, GO:0,036,293, GO:0,061,418) leukocyte activation involved in inflammatory responses (GO:0,002,758), the production of molecular mediators involved in inflammatory responses (GO:0,070,498), and inflammatory responses to antigen stimulation (GO:0,002,220). These results confirmed that DTT could treat ICH by reducing inflammatory reaction, reducing oxidative stress and inhibiting apoptosis (Fig. 8).

Fig. 8
figure 8

Pathway enrichment analysis of the targets of DTT

Potential mechanisms analysis of DTT in treating ICH

Pathological changes of ICH are related to many factors, among which oxidative stress and apoptosis are the key factors. When oxidative stress occurs, a large amount of reactive oxygen species will damage the blood–brain barrier, mediate demyelination and axonal injury, and activate various signal pathways to induce and aggravate autoimmune inflammation and neuronal apoptosis. We found 100 pathways shared by pathogenic genes, including MAPK signaling pathway (hsa04010), PI3K-Akt signaling pathway (HSA 04,151) and so on. More and more evidences show that these pathways and the pathogenesis or therapeutic targets of ICH, such as MAPK signaling pathway (hsa04010), mainly regulate protein synthesis, cell proliferation and apoptosis. MAPK plays a positive role in the initiation and recognition of oxidative stress, which has a positive correlation with brain injury. Using MAPK-related protein blockers can play a role in brain protection, and the effect of combining blockers is better. This suggests that multiple blocking of oxidative stress-related signaling pathways is one of the research ideas to reduce brain injury. DTT can play a role in pathological changes of cerebral hemorrhage by regulating apoptosis and oxidative stress pathway. In order to deduce the underlying mechanism of DTT on ICH, a comprehensive signal pathway was constructed at the system level. In order to figure out the location of DTT target on the pathway, the first three columns were defined as the upstream and the others were defined as the downstream of the pathway. We found that PI3K-Akt signaling pathway (hsa04151) is one of the most important pathways in the treatment of ICH with DTT (Fig. 9).

Fig. 9
figure 9

Distribution of ingredients of DTT on the compressed ICH pathway

Experimental validation in vitro

The effects of 6-Shogaol, 6-Singerol, kaempferol, and sitostrol on OGD model were evaluated. Compared with control group, the cell viability of OGD groups were significantly reduced. Compared with control group, the effect of cell viability was significantly decreased by 31.47% in the hypoxia treated cells (Fig. 10). However,6-Singerol (40 μM、80 μM、120 μM、160 μM and 200 μM) markedly increased the cell viability level by18.81%、19.98%、21.96%、16.53% and 23.47%.Compared with control group, the effect of cell viability was significantly decreased by 36.40% in the hypoxia treated cells. However,6-Singerol (160 μM and 200 μM) markedly increased the cell viability level by 20.96% and 20.44%.Compared with control group, the effect of cell viability was significantly decreased by 40.61% in the hypoxia treated cells. However, kaempferol (200 μM) markedly increased the cell viability level by 29.01%.Compared with control group, the effect of cell viability was significantly decreased by 35.08% in the hypoxia treated cells. However, sitostrol (200 μM) markedly increased the cell viability level by 23.65%.The above results demonstrated that 6-Shogaol, 6-Singerol, kaempferol, and sitostrol possessed protect effect in hypoxia treated HT22 cells.

Fig. 10
figure 10

Effects of 6-Shogaol (A),6-Singerol (B), kaempferol (C), and sitostrol (D) on cell viabilities. ***p < 0.001 compared with control group. #p < 0.05, #p < 0.05, ###p < 0.001 compared with the OGD group


Reducing non-pharmacological ingredients and improving curative effects are the main objectives of formula optimization. According to the theory of Chinese medicine, different herb medicines make up prescriptions, but whether the herbs or ingredients in prescriptions are necessary, especially for certain indications, needs to be analyzed and verified. By optimizing the formula, the medicinal materials or ingredients with certain efficacy were screened out, which made the formula more clarified and the efficacy more enhanced. In order to better capture the clinical efficacy of classical prescriptions, bioinformatics and systems pharmacology methods were combined to study the coverage rate of changed targets, and the relevant functional analysis was made on the coverage of pathogenic genes. The changed targets responds to the combination changes of different herbs and different chemical ingredients in each formula. In order to find the optimal FCIG, the optimization space definition and ingredient reverse search strategy were applied to evaluate it. At present, how to optimize and obtain FCIG and deduced underlying mechanisms is the foundation of TCM research. TCM analysis focus on the holistic view and regards the body as a whole. Systems pharmacology focuses on analyzing the action mechanisms of formula from a systematic and holistic perspective, which accords with the theoretical system of TCM research. Systems pharmacology emphasizes the multi-target regulation of multiple signal pathways to promote the synergistic effect of drugs and reduce toxic and side effects. At present, systems pharmacology has been widely used in the research of TCM formulas, especially to determine the molecular mechanism of treating complex diseases with TCM formula. However, there are few reports on optimization of TCM formulas based on systems pharmacology. On this basis, a comprehensive optimization strategy of DTT based on network pharmacology is proposed, and the key ingredients of DTT in treating ICH are obtained, and the potential mechanism of these ingredients is analyzed.

Our method has two advantages: 1. In the process of analyzing the treatment mechanism, network pharmacology has formed a fixed analysis rule, that is, according to the chemical properties of Chinese medicine ingredients, through screening, predicting targets and analyzing potential mechanisms of action. This flow chart solves the molecular mechanism of some prescriptions for treating complex diseases with TCM, but there is also a problem that there is no reference in the optimization space. In this paper, we identified the representative pathogenic genes of ICH, and these pathogenic genes have been reported in the literature. For example, APP, NOTCH3 weighted gene regulatory network of ICH, Zhu Bin et al. found that amyloid precursor protein (APP) mutation can cause typical pathological changes of AD and perivascular amyloid deposition. Fu Jiayu et al. showed that NOTCH3 gene mutation may cause cerebral hemorrhage by changing the structure and function of cerebral small vessels. We present a reverse optimization model based on the association between disease genes and ingredient targets is proposed, which provides space for optimization based on effective proteins. This method can well determine the optimization space of the target. Second, reverse searching related ingredients based on the optimized space provided by effective protein. The results showed that the enrichment functional pathway of effective protein could cover 96% of the enrichment functional pathway of disease genes. It is proved that our strategy of selecting effective proteins to construct the target optimization space is correct. On the basis of optimization space provided by effective protein, CI model was used to optimize the contribution degree, and finally 82kgec was optimized. The target of FCIG is closely related to the pathogenesis and functional annotation of ICH. This proves the reliability of our optimization space and CI model again.

At present, network pharmacology provides a powerful tool for exploring the compatibility and action mechanism of TCM prescriptions. However, there are some limitations. For example, more main active ingredients should be considered in animal medical research of DTT. We will verify the efficacy and mechanism of active ingredients in treating ICH through in vivo or in vitro experiments. The results show that the model has good accuracy in screening FCIG in TCM formula and provides reference for optimization and mechanism deduction of TCM formulas.


A network pharmacology model-based bioinformatics algorithm was established to obtain functional core ingredients group and decode the mechanisms of DTT in the treatment of ICH. Compared with other published work, potential effect space construction strategy based on novel node importance calculation method and validation strategy and maximum targeting weight model for mechanism speculation were reported. In addition, this new systemic pharmacology model closely combines the representative pathogenic genes of ICH, which can reflect the incidence of ICH well, and can effectively screen out the core ingredients of DTT in treating ICH.

Results of the in vitro validation that 6-Shogaol, 6-Singerol, kaempferol, and sitostrol possessed protect effect in hypoxia treated HT22 cells demonstrated that the core ingredient group selected by us based on system pharmacology had significant effect on ICH.

Our research is a computational mining work based on pharmacological basic data, which provides a feasible scheme to reduce the verification scale for the experiment, provides methodological reference for optimization of core ingredients group and interpretation of the molecular mechanism in the treatment of complex diseases using TCM. Our model has been proved to be effective in the compound optimization of DTT for the treatment of ICH. In future studies, we hope to apply this model to more studies and get better improvements, so as to provide methodological reference for the treatment of ICH with Chinese medicine.

Availability of data and materials

The datasets used or analysed during the current study are available from the corresponding author. Similarity Ensemble Approach: GeneCards database: BioGRID: HPRD:



Traditional Chinese medicine


Hypertensive cerebral hemorrhage


Potential effect space


Contribution index


Ditan decoction


Functional core ingredients group




Gene ontology


Protein–protein interactions


  1. Sun W, Liu WD. Pathogenesis of Hypertensive Intracerebral Hemorrhage. Foreign Me Sci(Cerebrovascular Diseases). 2005;10:756–8.

    Article  Google Scholar 

  2. Zeng J, Chen Y, Ding R, Feng L, Fu Z, Yang S, Deng X, Xie Z, Zheng S. Isoliquiritigenin alleviates early brain injury after experimental intracerebral hemorrhage via suppressing ROS- and/or NF-κB-mediated NLRP3 inflammasome activation by promoting Nrf2 antioxidant pathway. J Neuroinflammation. 2017;14(1):119.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  3. Feng L, Chen Y, Ding R, Fu Z, Yang S, Deng X, Zeng J. P2X7R blockade prevents NLRP3 inflammasome activation and brain injury in a rat model of intracerebral hemorrhage: involvement of peroxynitrite. J Neuroinflammation. 2015;12:190.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  4. Yang D, Han Y, Zhang J, Chopp M, Seyfried DM. Statins Enhance Expression of Growth Factors and Activate the PI3K/Akt-mediated Signaling Pathway after Experimental Intracerebral Hemorrhage. World J Neurosci. 2012;2(2):74–80.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  5. Han Y, Seyfried D, Meng Y, Yang D, Schultz L, Chopp M, Seyfried D. Multipotent mesenchymal stromal cell-derived exosomes improve functional recovery after experimental intracerebral hemorrhage in the rat. J Neurosurg. 2018;131(1):290–300.

    Article  PubMed  Google Scholar 

  6. Ling, Zhou LF, Huang DH, Wang HQ, Li QW. Tongfu Xingshen Decoction in the treatment of 65 cases of hypertensive cerebral hemorrhage of qi deficiency and blood stasis type. Zhejiang J Tradit Chin Med. 2019;(10):730–1.

  7. Guo SR, Lai ZY, C YgJ, H HZ. Research on the application of Angong Niuhuang Pill in different types of hypertensive intracerebral hemorrhage. Clin J Tradit Chin Med. 2014;(06):564–6.

  8. Zhou YF, Wang JG. Effect of Huoxue Ditan Decoction on Neural Function Recovery in Hypertensive Cerebral Hemorrhage. World Latest Med Inf. 2017;(19):91.

  9. Xenarios I, Rice DW, Salwinski L, Baron MK, Marcotte EM, Eisenberg D. DIP: the database of interacting proteins. Nucleic Acids Res. 2000;28(1):289–91.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  10. Pinkert S, Schultz J, Reichardt J. Protein interaction networks–more than mere modules. PLoS Comput Biol. 2010;6(1):e1000659.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  11. Oughtred R, Rust J, Chang C, Breitkreutz BJ, Stark C, Willems A, Boucher L, Leung G, Kolas N, Zhang F, Dolma S, Coulombe-Huntington J, Chatr-Aryamontri A, Dolinski K, Tyers M. The BioGRID database: A comprehensive biomedical resource of curated protein, genetic, and chemical interactions. Protein Sci : a publication of the Protein Society. 2021;30(1):187–200.

    Article  CAS  Google Scholar 

  12. von Mering C, Huynen M, Jaeggi D, Schmidt S, Bork P, Snel B. STRING: a database of predicted functional associations between proteins. Nucleic Acids Res. 2003;31(1):258–61.

    Article  CAS  Google Scholar 

  13. Fabregat A, Korninger F, Viteri G, Sidiropoulos K, Marin-Garcia P, Ping P, Wu G, Stein L, D’Eustachio P, Hermjakob H. Reactome graph database: Efficient access to complex pathway data. PLoS Comput Biol. 2018;14(1):e1005968.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  14. Zanzoni A, Montecchi-Palazzi L, Quondam M, Ausiello G, Helmer-Citterich M, Cesareni G. MINT: a Molecular INTeraction database. FEBS Lett. 2002;513(1):135–40.

    Article  CAS  PubMed  Google Scholar 

  15. Aranda B, Achuthan P, Alam-Faruque Y, Armean I, Bridge A, Derow C, Feuermann M, Ghanbarian AT, Kerrien S, Khadake J, Kerssemakers J, Leroy C, Menden M, Michaut M, Montecchi-Palazzi L, Neuhauser SN, Orchard S, Perreau V, Roechert B, van Eijk K, Hermjakob H. The IntAct molecular interaction database in 2010. Nucleic Acids Res. 2010;38((Database issue)):D525–31.

    Article  CAS  PubMed  Google Scholar 

  16. Tao W, Xu X, Wang X, Li B, Wang Y, Li Y, Yang L. Network pharmacology-based prediction of the active ingredients and potential targets of Chinese herbal Radix Curcumae formula for application to cardiovascular disease. J Ethnopharmacol. 2013;145(1):1–10.

    Article  CAS  PubMed  Google Scholar 

  17. Yin C, Xiao X, Balaban V, Kandel ME, Lee YJ, Popescu G, Bogdan P. Network science characteristics of brain-derived neuronal cultures deciphered from quantitative phase imaging data. Sci Rep. 2020;10(1):15078.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  18. Maetschke SR, Yuan Z. Exploiting structural and topological information to improve prediction of RNA-protein binding sites. BMC Bioinformatics. 2009;10:341.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  19. Haw DJ, Pung R, Read JM, Riley S. Strong spatial embedding of social networks generates nonstandard epidemic dynamics independent of degree distribution and clustering. Proc Natl Acad Sci USA. 2020;117(38):23636–42.

    Article  PubMed  PubMed Central  Google Scholar 

  20. Keiser MJ, Roth BL, Armbruster BN, Ernsberger P, Irwin JJ, Shoichet BK. Relating protein pharmacology by ligand chemistry. Nat Biotechnol. 2007;25(2):197–206.

    Article  CAS  PubMed  Google Scholar 

  21. Chen YH, Zhao XX, Wang WQ, Duan TX. Simultaneous Determination of Glycyrrhizin, Liquiritin and Isoliquritigenin in Licorice by HPLC. Chin J Inf Tradit Chin Med. 2009;8:52–4.

  22. Zhang WK, Song S, Cui XB, Xu ZY. Determination of 6-Gingerol and 6-Shogaol in Ginger of Different Districts in China. Chin Pharm J. 2009;22:1692–4.

  23. Liu YC, Zhang J. Simultaneous determination of hesperidin and naringin in orange tangerine loquat capsule by HPLC.  Anhui Med Pharm J. 2013;12:2044–5.

  24. Zhang J, Xu YC, Lin D, Yuan XL. Determination of four chemical components in rhizoma arisaematis from different origins by multi-assay. J Chin Med Mater. 2020;01:141–4.

  25. Shang RX, Xu WD, Xu XQ, Peng Y. Analysis of the major metabolites of Sijunzi decoction in rat serum based on HPLC-MS. J Jiangsu Univ (Med Ed). 2018;6:529–33.

  26. Chen YQ, Liu H, Zhang J. Determination of Uridine, Guanosine and Adenosine in Pinellia ternate by RP-HPLC. Her Med. 2013;12:1634–6.

  27. Peng C, Yu SL, Zhang J, Ji WP, Qiang W, Li Z, Zhou QW. Simultaneous determination of four triterpenoid acids in Poria by HPLC. J Chin Med Mater. 2017;7:1643–6.

  28. Zhan ZB, Deng KZ, Xiong Y, Ding BQ, Deng MZ. Simultaneous determination of six ingredients in Aurantii Fructus Immaturus by HPLC. Chin J Hosp Pharm. 2015;12:1080–2.

  29. Li SH, Zhao R, Zhao SH, Xing YP, Chen P. Research of Polysaccharide and Flavonoids Content in Bamboo Shavings by Different Processing Methods. J Liaoning Univ Tradit Chin Med. 2017;9:59–62.

  30. Guan D, Shao J, Deng Y, Wang P, Zhao Z, Liang Y, Wang J, Yan B. CMGRN: a web server for constructing multilevel gene regulatory networks using ChIP-seq and gene expression data. Bioinformatics (Oxford, England). 2014;30(8):1190–2.

    Article  CAS  PubMed  Google Scholar 

  31. Guan D, Shao J, Zhao Z, Wang P, Qin J, Deng Y, Boheler KR, Wang J, Yan B. PTHGRN: unraveling post-translational hierarchical gene regulatory networks using PPI ChIP-seq and gene expression data. Nucleic Acids Res. 2014;42(Web Server issue):W130–6.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  32. Zhu B, Chen J, Q HF, Deng RD, Li YW, Li H, Chen DF.Early Changes of Pathology and Cognitive Behavior in APP/PS1 Transgenic Mice with Alzheimer's Disease. J Guangzhou Univ Tradit Chin Med. 2012;(02):193–196+226.

  33. Fu JY, Tang L, Fan DS. Genetic variants in NOTCH3 associated with spontaneous intracerebral hemorrhage. Chin J Front Med Sci (Electron Version). 2019;(10):11–8.

  34. Liu HQ, Guo SH, Shen YS, Li XM. Study in antioxidative effect of β-sitosteroll. Acad J Guangdong College Pharmacy. 2004;03:281–3.,2004.03.033.

    Article  Google Scholar 

  35. Bai X, Zhang X, Chen L, Zhang J, Zhang L, Zhao X, et al. The neuroprotective mechanism of naringenin on rats with permanent cerebral ischemia: down-regulating the expression of NOD2,RIP2,NF-KB,MMP-9, up-regulating the expression of Claudin-5. (eds.) Chin Med Assoc. 2015.

  36. Zeng GF, Zong SH, Fu SW, Nong MN, Li KK, Yan FN. Protective Effect of 6-Gingerol on PC12 Cells Induced by Amyloid β-Protein by PI3K/Akt/GSK-3βSignaling Pathway. Nat Prod Res Dev. 2015;11:1862–5.

    Article  CAS  Google Scholar 

  37. Wang XL, Ai X, Guo JK, Wei W, Liu YH, Yuan Z, Zhang CM. The Effect of α-linolenic acid on acute inflammation and neurological functional recovery of mice with traumatic brain injury. Chin J Nerv Ment Dis. 2018;(05):294–8.

  38. Shi J, Gao Z, Ke K. Effects of total saponins of panax ginseng on the proliferation of endogenous neural stem cells following intracerebral hemorrhage. J Nantong Univ (Med Sci). 2009;(04):253–5.

  39. Lampiasi N, Montana G. The molecular events behind ferulic acid mediated modulation of IL-6 expression in LPS-activated Raw 264.7 cells. Immunobiology. 2016;221(3):486–93.

    Article  CAS  PubMed  Google Scholar 

  40. Liu W, Zhang Y, Guo JC, Lian PY, W MH, Ding WT. The protective effect of naringin on cerebral ischemia reperfusion injury. J Apoplexy Nerv Dis. 2017;(04):292–4.

  41. Sun Cm, Xie DX, Zheng WH. Protective effect of 6-gingerol on cerebral ischemia injury model mice. Shandong Med J. 2011;(52):41–43+136.

  42. Wang S, Sun Y, Li CM, Lu Q. Research Progress of Stigmasterol. China Pharm. 2019;(23):96–8.

  43. Wang L, Q H, Bao TC, Tao T, Hou JL, Xiong XG, Xie Y. Effect of Ditan Decoction on Expression of Activated Caspase-3 around Hematoma after Intracerebral Hemorrhage in Rats. J Tradit Chin Med Univ Hunan. 2005;(06):19–21.

  44. Wang L, Liang Q, Wang T, Bao T, Hou J. Effects of ditan decoction on the release of Cytochrome C from neuronal mitochondria in rats with intracerebral hemorrhage. Chin J Clin Rehabil. 2005;(25):61–63+261.

Download references


Not applicable.


This study is financially supported by the Startup fund from Scientific Research Project of Guangdong Provincial Administration of Traditional Chinese Medicine (No. 20181111), the Startup fund from Guangzhou Mulicipal Science and Technology Bureau (No. 202002030221), the Natural Science Foundation Council of China [grant No. 31501080],the 17th Bajian Talent Project of Guangdong Provincial Hospital of Traditional Chinese Medicine (No. BJ2022YL06),the 13th Chaoyang Talent Project of Guangdong Provincial Hospital of Traditional Chinese Medicine(ZY2022YL28).

Author information

Authors and Affiliations



DJX, ZZ and KXW provided the concept and designed the study, DJX, ZZ, ZZP and YBS conducted the analyses and wrote the manuscript. DGG, TH, NZ, DJX participated in data analysis. ZZ and YPC carried out experimental validation. DGG and ZZ contributed to revising and proof-reading the manuscript. All authors read and approved the final manuscript.

Corresponding authors

Correspondence to Ke-xin Wang, Dao-gang Guan or Tao Huang.

Ethics declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no conflict of interest.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Additional file 1:

Supplementary Table 1. Protein-Protein Interaction Network.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit The Creative Commons Public Domain Dedication waiver ( applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Zhen, Z., Xue, Dj., Chen, Yp. et al. Decoding the underlying mechanisms of Di-Tan-Decoction in treating intracerebral hemorrhage based on network pharmacology. BMC Complement Med Ther 23, 44 (2023).

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: