In this paper, we propose an autonomous robot packing system named RoboPacker designed to tightly store cluttered general objects into shipping boxes with high space utilization, which is a fundamental process in numerous industrial applications. However, achieving tight packaging for general objects often demands significant labor from human packers, particularly in high-throughput scenes. Compared to existing robot packing approaches, RoboPacker effectively overcomes challenges such as diverse object appearances, severe occlusion, and crowded packing spaces. Specifically, we propose an open-vocabulary shape estimation method to reconstruct complete point clouds for cluttered objects. We also design effective interactions with object clutter to gather informative visual clues for shape estimation under high uncertainty. Additionally, we introduce a hierarchical reinforcement learning framework to optimize packing order, location, and orientation for maximum space utilization. The robotic packing system integrates these techniques with feasible manipulation methods for real-world implementation. In this way, RoboPacker achieves efficient packing of novel and irregular objects, which is more suitable for real deployment environments. The Real-world experiments demonstrate RoboPacker can tightly pack 20 densely cluttered everyday objects from 8 seen and 4 novel classes into the 40x40x20 cm shipping box with a 73.3% success rate.

Object packing by autonomous robots is an im-portant challenge in warehouses and logistics industry. Most conventional data-driven packing planning approaches focus on regular cuboid packing, which are usually heuristic and limit the practical use in realistic applications with everyday objects. In this paper, we propose a deep hierarchical reinforcement learning approach to simultaneously plan packing sequence and placement for irregular object packing. Specifically, the top manager network infers packing sequence from six principal view heightmaps of all objects, and then the bottom worker network receives heightmaps of the next object to predict the placement position and orientation. The two networks are trained hierarchically in a self-supervised Q-Learning framework, where the rewards are provided by the packing results based on the top height , object volume and placement stability in the box. The framework repeats sequence and placement planning iteratively until all objects have been packed into the box or no space is remained for unpacked items. We compare our approach with existing robotic packing methods for irregular objects in a physics simulator. Experiments show that our approach can pack more objects with less time cost than the state-of-the-art packing methods of irregular objects. We also implement our packing plan with a robotic manipulator to show the generalization ability in the real world.

Grasping in dense clutter is a fundamental skill for autonomous robots. However, the crowdedness and occlusions in the cluttered scenario cause significant difficulties to generate valid grasp poses without collisions, which results in low efficiency and high failure rates. To address these, we present a generic framework called GE-Grasp for robotic motion planning in dense clutter, where we leverage diverse action primitives for occluded object removal and present the generator-evaluator architecture to avoid spatial collisions. Therefore, our GE-Grasp is capable of grasping objects in dense clutter efficiently with promising success rates. Specifically, we define three action primitives: target-oriented grasping for target capturing, pushing, and nontarget-oriented grasping to reduce the crowdedness and occlusions. The generators effectively provide various action candidates referring to the spatial information. Meanwhile, the evaluators assess the selected action primitive candidates, where the optimal action is implemented by the robot. Extensive experiments in simulated and real-world environments show that our approach outperforms the state-of-the-art methods of grasping in clutter with respect to motion efficiency and success rates. Moreover, we achieve comparable performance in the real world as that in the simulation environment, which indicates the strong generalization ability of our GE-Grasp. Supplementary material is available at: https://github.com/CaptainWuDaoKou/GE-Grasp.

Aim: To investigate whether genotypes of XDH, GMPS and MOCOS were associated with azathioprine-induced adverse drug reaction (ADR) and had the gene-gene interactions with NUDT15 rs116855232 to induce leukopenia. Methods: Patients who had taken azathioprine were recruited. Genotyping of those gene was performed. Risk factor to ADR was analyzed by logistic regression. The generalized multifactor dimensionality reduction (GMDR) was assessed based on gene-gene interactions with ADR. Results: A total of 111 patients were included in this study, all of whom were Han Chinese. XDH rs2295475 was a risk factor of myelotoxicity (p = 0.022). NUDT15 rs116855232 was a risk factor of myelotoxicity, grade ≥2 leukopenia and drug treatment termination (p-values were <0.05). Rs2295475 and rs116855232 had a gene-gene interaction. The model was associated with grade ≥2 leukopenia (OR: 17.99; 95% CI: 4.11-78.81). Conclusion: Combined testing genotype for rs2295475 and rs116855232 could improve the prediction of azathioprine-induced leukopenia.

Cardiovascular complications of patients with type 2 diabetes mellitus (T2DM) threaten the health and life of numerous individuals. Recently, growth factor receptor-binding protein 10 (GRB10) was found to play a pivotal role in vascular complications of T2DM, which participates in the regulation of lipid metabolism of T2DM patients. The genetic variation of GRB10 rs1800504 is closely related to the risk of coronary heart disease in patients with T2DM. The development of GRB10 as a key mediator in the association of lipid metabolism with cardiovascular complications in T2DM is detailed in and may provide new potential concerns for the study of cardiovascular complications in T2DM patients.

目的 分析临床药师对心内科不合理医嘱的干预情况,促进合理用药.方法 对珠海市人民医院2019年1～12月心内科住院医嘱审核中临床药师的干预记录进行统计分析.结果 不合理医嘱干预记录共146例,干预成功率91.1％.不合理医嘱类型前3位分别是用药剂量不适宜、禁忌证用药、药物联用不适宜,分别占27.40％、24.66％、10.96％.结论 临床药师通过医嘱审核能促进合理用药,体现药师价值.

目的 研究黄嘌呤脱氢酶(xanthine dehydrogenase,XDH)和鸟苷酸合成酶(guanosine monophosphate synthetase,GMPS)基因多态性对硫唑嘌呤(azathioprine,AZA)所致骨髓抑制和脱发的影响,探讨上述位点联合NUDT15 rs116855232检测对预测AZA所致骨髓抑制的价值.方法 回顾性分析曾经服用或正在服用AZA治疗自身免疫性疾病患者,收集静脉血采用直接测序法测定GMPS rs61750368、rs61750369,XDH rs2295475、rs1884725、rs17011368、rs566362和NUDT15rs116855232基因型.依据不良反应分为2组,分析上述基因型与AZA所致骨髓抑制和脱发的相关性,NUDT15 rs116855232作为对照位点.结果 共纳入80例患者,2组在年龄、日剂量、疗程、BMI以及6-硫鸟嘌呤核苷酸浓度均无显著性差异.Logistic回归分析表明,AZA所致骨髓抑制与XDH rs2295475突变相关(P=0.003,OR=3.10,95％CI: 1.45～6.25),该基因预测骨髓抑制敏感度为69.6％,特异度为63.2％,ROC曲线AUC为0.66;NUDT15 rs116855232突变也和AZA所致骨髓抑制相关(P=0.008,OR=3.46,95％CI: 1.21～9.93);其余位点基因型暂未发现与骨髓抑制和脱发相关.rs116855232野生型且rs2295475突变型的患者对比两位点野生型的患者,其骨髓抑制的危险值升高(OR=6.53,P=0.004),而两位点同时突变者骨髓抑制的危险值更高(OR=51.67,P＜0.001).结论 在中国自身免疫性疾病患者中XDH rs2295475突变可能为AZA诱导骨髓抑制的独立危险因素,尽管预测准确度较低但仍有预测价值.尤其XDH联合NUDT15 rs116855232监测,可能会弥补NUDT15野生型者对骨髓抑制预测不足的风险,提高NUDT15突变型对骨髓抑制预测的准确性,但以上结果还需更多临床研究验证.

目的 调查某院急性冠脉综合征(ACS)患者发病后6个月的低密度脂蛋白(LDL-C)控制情况,为临床药师参与这类患者的管理提供参考.方法 筛选2017年9月～2018年6月某院收治且后续在该院规律复诊的ACS患者共73例,回顾性分析其6个月的LDL-C控制情况.结果78.1%的患者3个月内至少有监测1次LDL-C,达标率28.1%;21.9%的患者在出院后4～6个月才首次监测LDL-C,4～6个月共计28例患者监测LDL-C,达标率为25%.35.6%的患者出院时处方高强度他汀,64.4%的患者出院时处方中等强度他汀.结论 某院对ACS患者的血脂监测周期有待进一步规范,LDL-C达标率有待进一步提高,临床药师应积极参与此类患者的管理.

目的 了解高尿酸血症(HUA)患者降尿酸药物的使用现状,为合理用药提供参考.方法 通过HIS系统抽取2018年1-6月份三甲医院住院患者中有HUA诊断的病例230份,分析其降尿酸药物的使用合理性.结果 230例患者中有98例有降尿酸指征,其中仅35.71％的患者开启了降尿酸治疗;医师对降尿酸药物的使用存在较多不合理情况,主要表现在用法用量不适宜、适应证不适宜、违反禁忌证用药、用药监测不到位等情况.结论 医师应当加强对HUA患者的评估,对有降尿酸指征的患者尽早开启降尿酸治疗.医院应加强降尿酸药物的管理和宣传,以提高医师降尿酸药物的合理使用水平.

目的 了解珠海地区居民合理用药常识和用药行为现状.方法 在2016年10月至2017年4月采用问卷方式对珠海地区居民进行调查,内容包括一般情况、用药习惯、用药依从性、合理用药认识、药物不良反应及用药错误的发生情况等.结果 部分被调查者有不正确的用药习惯.被调查者合理用药常识普遍缺乏,用药依从性不尽人意.中老年人(≥45岁)发生药品不良反应的风险高于青年人(<45岁)(P<0.05)."自我判断"购药者更容易服用不恰当的药物(P<0.05)."经常服药前阅读说明书"者更少出现用法用量不正确的情况(P<0.05)."经常服药前留意药品有效期"者更少发生服用过期药品的事件(P<0.05).结论 珠海地区居民用药习惯有待改善,合理用药认识水平亟待提高.药师应加强社区药学服务,推动合理用药公众教育.

Fibroblast activation protein- (FAP) is a promising tumor-associated target expressed by reactive stromal fibroblasts in tumor tissue. FAP has a postprolyl peptidase activity and can specifically cleave N-terminal benzyloxycarbonyl (Z)-blocked peptides, such as the substrate Z-Gly-Pro-AMC. Doxorubicin (DOX) is an effective antitumor drug, but its application is greatly limited by toxic adverse effects owing to poor tumor selectivity. Based on these facts, we previously designed a FAP-targeting prodrug of doxorubicin (FTPD) which can be selectively hydrolyzed by FAP. FTPD can retain potent antitumor efficacy and has favorable tumor targeting. The present study aimed to further evaluate the toxicological profile and the safety pharmacological property of FTPD in vitro and in vivo. The cytotoxicity assay showed that FTPD displayed markedly lower cytotoxicity to 3T3 cells and HEK-293 cells compared with DOX. In the short-term toxicity study, mice treated with 25mg/kg of FTPD showed no obvious change in the appearance and general behavior, and no case of mortality was observed within 14 days. Unlike DOX, FTPD exhibited reduced toxicity to heart, liver, kidney, spleen as well as peripheral white blood cells in mice. Moreover, open file test and general pharmacology study were also conducted correspondingly in mice and beagle dogs. It was found that FTPD may not produce significant pharmacological effects on spontaneous locomotor activity and cardiovascular-respiratory system except for a transient decreasing in systolic blood pressure. Taken together, the results of this work suggest that FTPD has more favorable toxicological profile and better drug safety compared with its parent drug DOX.

OBJECTIVE:To explore the intrinsic connection between activation of classical nuclear factor-κB (NF-κB) pathway and gefitinib resistance in human lung adenocarcinoma H1650 cells.METHODS:Human lung adenocarcinoma H1650 cells were exposed to gefitinib continuously for 60 days to obtain resistant H1650 cells. The expressions of P-IκBα, P-p50 and P-p65 in the cytoplasm or nuclei were detected using Western blotting in human lung adenocarcinoma HCC827 cells, parental H1650 cells and gefitinib-resistant H1650 cells. The effects of gefitinib alone or in combination with PDTC on the survival rate and expressions of NF-κB P-p50 and P-p65 were compared among the 3 cell lines.RESULTS:Gefitinib-resistant H1650 cells showed increased cytoplasmic and nuclear P-IκBα expressions. The expressions of P-p50 and P-p65 differed significantly among the 3 cell line, decreasing in the order of resistant H1650 cells, parental H1650 cells, and gefitinib sensitive HCC827 cell lines (P<0.05 or 0.01). Treatment with gefitinib alone resulted in a significantly lower cell inhibition rate in resistant H1650 cells than in the parental H1650 cells (P<0.05) and HCC827 cells (P<0.01). The resistant H1650 cells had a significantly higher expression of P-p50 and P-p65 than other two cell lines (P<0.05). In both the resistant and parental H1650 cells, gefitinib significantly lowered P-p50 and P-p65 expressions (P<0.05 or 0.01), and the combined treatment with gefitinib and PDTC significantly decreased the cell survival rate and further lowered the cytoplasmic and nuclear expressions of P-p50 and P-p65 (P<0.01 or 0.01).CONCLUSION:The activation of classical NF-κB pathway is a key factor contributing to transformation of the parental H1650 cells into gefitinib-resistant cells. Gefitinib combined with PDTC can inhibit P-IκBα production and NF-κB P-p50 and P-p65 activation to suppress the survival of residual H1650 cells and the generation of gefitinib-resistant cells.

病例:患儿,女,9 岁 9 个月.2017 年 11 月 17 日无明显诱因下出现阵发性头痛,无恶心、呕吐,无畏寒、高热,无意识障碍,无行走不稳,无肢体麻木,无抽搐.2017 年11 月 19 日于我院门诊就诊,行颅脑磁共振平扫示:左侧小脑半球异常信号灶,考虑可能为脑血肿,脑血管畸形伴出血,为进一步治疗于 2017 年 11 月 22 日收入我院神经外科.

Fibroblast activation protein-α (FAPα) is a serine protease of the post-prolyl peptidase family that is specifically expressed in the majority of human epithelial tumors, but not in normal tissues. In this study, we demonstrated the anti-tumor activity of a novel targeting drug formed by conjugating epirubicin (EPI) with an FAPα-specific dipeptide (Z-Gly-Pro) and named it Z-GP-EPI. Consistent with this tumor-targeting delivery strategy, the results illustrated that Z-GP-EPI could release EPI efficiently after incubating with FAPα and could exhibit similar antitumor effects as EPI in vitro in FAPα over-expressed tumor cells (4T1/FAPα+). Furthermore, the evaluation of antitumor activity of Z-GP-EPI in vivo was implemented in a 4T1/FAPα+ tumor-bearing mice xenograft model. Our results illustrated that Z-GP-EPI had similar antitumor effects in 4T1/FAPα+ tumor-bearing mice and showed no visible cardiotoxicity side effects compared with free EPI. Thus, our study indicated that this FAPα-activated prodrug targeting strategy may provide a new mechanism for the targeted delivery of antitumor agents and improve their safety levels.

OBJECTIVE:In this study, a systematic evaluation was conducted to estimate the efficacy and safety of ticagrelor for treating acute coronary syndrome (ACS) in general ACS patients and a diabetes mellitus (DM) group. METHODS:A search of PubMed, Cochrane Central Register of Controlled Trials, Web of Science, CNKI databases was conducted to analyze relevant randomized controlled trails (RCTs) of ticagrelor treating ACS during 2007 to 2015. Article screening, quality accessing and data extracting was independently undertaken by two reviewers. A meta-analysis was performed to clarify the efficacy and safety of ticagrelor in general ACS patients, and a meta-regression analysis was taken to demonstrate the efficacy and safety of ticagrelor in DM patients compared with general ACS patients. RESULT:Twenty-two studies with 35004 participants were included. The meta-analysis result implicated that ticagrelor could: 1) reduce the incidence of the composite endpoint [OR = 0.83, 95%CI (0.77, 0.90), P<0.00001] and the incidence of myocardial infarction [OR = 0.81, 95%CI (0.74, 0.89), P = 0.0001]; 2) not statistically reduce the incidence of cardiovascular death, the incidence of stroke and the incidence of bleeding events; 3) increase the incidence of dyspnea [OR = 1.90, 95%CI (1.73, 2.08), P<0.00001] compared with clopidogrel. Meanwhile, compared with prasugrel, ticagrelor could 1) reduce the platelet reactivity of patients at maintenance dose [MD = -44.59, 95%CI (-59.16, -30.02), P<0.00001]; 2) not statistically reduce the incidence of cardiovascular death, the platelet reactivity of patients 6 hours or 8 hours after administration, or the incidence of bleeding events; 3) induce the incidence of dyspnea [OR = 13.99, 95%CI (2.58, 75.92), P = 0.002]. Furthermore, the result of meta-regression analysis implicated that there was a positive correlation between DM patients and the platelet reactivity of patients 6 hours and 8 hours after administration, but there was no obvious correlation between DM patients and general ACS patients in other endpoints. CONCLUSION:Ticagrelor could reduce the incidence of composite endpoint of cardiovascular death, myocardial infarction and stroke as well as platelet reactivity in DM patients with ACS, while not increasing the risk of bleeding. Because there are differences in platelet reactivity between DM patients and general ACS patients, we suggest that caution is needed when using ticagrelor in clinical applications.

Objective To study the FMS-like tyrosine kinase-3 (FLT3) gene, NPM1 gene and c-kit gene mutations in acute myeloid leukemia (AML) by extracting DNA from the storage of bone marrow slides, and to investigate the relationship between the three gene mutations and clinical features in AML. Methods The bone marrow slides of 55 patients diagnosed with AML were enrolled in this study. The PCR, DNA sequencing and molecular cloning were used to detect and analyse the FLT3-ITD, NPM1 and c-kit gene mutations. Patients' remission, progression and survival time were also recorded. Results The DNA was successfully extracted from the bone marrow slides with -20 ℃ frozen storage without Wright stained, chemically fixed, and room temperature storage Wright stained discoloured by phenol ∶ chloroform ∶ isoamyl alcohol method, which can be used in PCR, direct sequencing and molecular cloning sequencing analysis. 10 of the 55 cases (18.2 %) were FLT3-ITD positive, including 9 cases with heterozygous mutations and 1 case with homozygous mutation. FLT3-ITD positive group had lower complete remission (CR) rate, shorter event-free survival (EFS) time and overall survival (OS) time than the negative group (P< 0.05). 9 of the 55 cases (16.4 %) had NPM1 heterozygous gene mutations, all belonging to type A. The EFS rate of the patients with NPM1 mutation was higher in 10 months and the OS rate was higher in 19 months (P< 0.05). 3 of 9 NPM1 mutations patients were FLT3-ITD positive. The CR rates of the four groups after initial remission induction therapy in order were NPM1+FLT3-ITD-, NPM1-FLT3-ITD-, NPM1-FLT3-ITD+, NPM1+FLT3-ITD+(P<0.05). Besides, NPM1-FLT3-ITD+was a risk factor affecting the OS (RR=1.250, P=0.005). 2 of the 55 cases (3.6 %) had c-kit gene mutations, namely mutant D816H and mutant D816V. The c-kit gene mutations were not found in patients with FLT3-ITD and NPM1 mutations. Conclusions The FLT3-ITD mutation is a poor prognosis molecular marker in AML, and NPM1 mutation is a good factor for the prognosis. NPM1-FLT3-ITD+is a risk factor affecting OS. The rate of c-kit gene mutation is low in AML, without the overlap of FLT3 and NPM1 mutations.

OBJECTIVE:To investigate and evaluate the diagnostic and predictive value of procalcitonin(PCT)for ICU fever patients with bloodstream infections(BSI). METHODS:Medical records of 233 ICU fever patients in our hospital during Mar. 2012 to Dec. 2014 were analyzed retrospectively. Serum levels of PCT and CRP were compared among patients with different blood cul-ture results,pathogen types and species of bacterial infection. The diagnostic and predictive value of PCT and CRP for BSI were evaluated and compared by using ROC curves. RESULTS:Among 233 fever patients,there were 74 cases of positive blood culture with positive rate of 31.76%. The serum levels of PCR and CRP in patients with positive blood culture were higher than in those with negative blood culture and contamination,with statistical significance(P<0.001). There was no statistical significance in se-rum levels of PCR and CRP between negative blood culture group and contamination group(P>0.05). Among 74 patients with pos-itive blood culture,there were 45 cases of gram-negative bacterial infection,accounting for 60.81%;18 cases of gram-positive bac-terial infection,accounting for 24.32%;11 cases of fungi infection,accounting for 14.86%. Serum level of PCT in patients with gram-negative bacterial infection was higher than in those with gram-positive bacterial infection and fungi infection,and serum lev-el of PCT in patients with gram-positive bacterial infection was higher than in those with fungi infection,with statistical signifi-cance(P<0.05). There was no statistical significance in serum level of CRP among patients with 3 kinds of bacterial infection(P>0.05). Main bacteria of infection in fever patients included Acinetobacter,Staphylococcus aureus,Pseudomonas aeruginosa,Esche-richia coli and Klebsiella pneumoniae;there was statistical significance in serum level of PCT among various bacterial infection (P<0.05);there was no statistical significance in serum level of CRP (P>0.05). ROC curve analysis demonstrated there was statistical significance in differentiation of distinguishing positive and negative blood culture by PCT and CRP (P<0.001);AUROC were 0.789 [95%CI (0.732, 0.845)] and 0.629 [95%CI(0.568,0.690)];cutoff value were 1.2 ng/ml and 81.4 mg/L. There was statistical significance in the differentiation of distinguishing positive blood culture and contamination by PCT and CRP (P<0.001);AUROC were 0.805 [95%CI(0.711, 0.899)] and 0.673 [95%CI(0.540,0.805)];cutoff value were 0.5 ng/ml and 73.4 mg/L. CONCLUSIONS:PCT may be a useful tool for the diagnosis of BSI and the judgment of blood culture results in ICU fever patients,and its predictive function was better than CRP. Serum level monitoring of PCT can guide the early empirical anti-infective therapy of BSI.

目的：观察晚期非小细胞肺癌靶向治疗进展后，比较化疗、姑息及原靶向3种治疗方式的疗效，探讨3种治疗方式对患者进展及生存时间的影响。同时，评估患者进展模式，为个体化治疗晚期非小细胞肺癌提供理论及实践依据。方法：收集50例ⅢB ～Ⅳ期EGFR突变的非小细胞肺癌患者，一线治疗均为靶向治疗，二线采用不同治疗方式，计算肿瘤进展时间及总生存时间。结果：靶向治疗主要的进展部位依次为肺、中枢神经系统、骨等；化疗组、姑息组及原靶向组平均肿瘤进展时间分别为10．67、5．92及5．88个月，3组肿瘤进展时间比较，差异无统计学意义（P ＞0．05）；化疗组、姑息组及原靶向组平均生存时间为20．04、7．53及14．21个月，化疗组总生存时间最长，姑息组最短，原靶向组介于两者之间，3组总生存时间差异有统计学意（P ＜0．05）。结论：晚期非小细胞肺癌靶向治疗进展后，肺为主要进展部位；化疗可使患者临床获益，总生存时间延长。

Purpose : To systematically review the clinical efficacy and adverse reactions of Paclitaxel for the treatment of mammary cancer. Math : We searched Web of knowledge, PubMed, VIP information and CNKI (to October 2013) on randomised controlled trial about Paclitaxel for the treatment of mammary cancer and retrieved relevant reference and research material by hand. Two authors independently screened document, extracted data and assessed the quality according to inclusion and exclusion criteria, we finally used the software RevMan 5.2 from Cochrane for Meta-analysis. Result : 18 randomized controlled clinical study were brought into our study according to inclusion and exclusion criteria, including 10712 patients. The result of meta-analysis showed that the odds ratios of Paclitaxel for adjunctive therapy [OR = 1.64, 95% CI (1.40, 1.92), P <0.00001] was better than conventional drugs, while the overall survival was no significant difference between Paclitaxel and conventional drugs. The further Subgroup analysis showed that the efficacy of Paclitaxel for adjunctive therapy was better than cyclophosphamide [OR = 1.41, 95%C I (1. 07, 1.85), P =0. 01] and NVB [OR = 2.10, 95% CI (1.33, 3.30), P =0.001]. The adverse reactions analysis results showed the ratio of myelosuppression and alopecia by treated with Paclitaxel was improved, while the occurrence of gastrointestinal reaction rate was decreased. Conclusion : The current evidence showed Paclitaxel was effective for the adjuvant treatment of breast cancer, but the above conclusions still need future expansion of more samples, high quality RCT verify.

Structure activity relationship (SAR) of fibroblast activation protein alpha (FAP) inhibitors will be useful to evaluate bioactivities of candidates. To discuss SAR of FAP inhibitors, two alignment styles were carried out to build QSAR models of FAP inhibitors. HQSAR was used to construct 2D-QSAR after the selection of training set and test set by principal component analysis method. Meanwhile, 3D-QSAR models were constructed by comparative molecular field analysis and comparative molecular similarity indices analysis method and optimized by FOCUS method. All the QSAR models were validated by cross-validation and test set, and the targeted QSAR model was selected by comprehensive evaluation containing cross-validation coefficient, correlation coefficient and consistency with docking studies. The result suggests that 2D-QSAR model may be insufficient to evaluate SAR of FAP inhibitors, while 3D-QSAR model with S+H+D_F functional fields could be applied to characterize the SAR based on docking conformation alignment.