1.ResNet-Vision Transformer based MRI-endoscopy fusion model for predicting treatment response to neoadjuvant chemoradiotherapy in locally advanced rectal cancer: A multicenter study.
Junhao ZHANG ; Ruiqing LIU ; Di HAO ; Guangye TIAN ; Shiwei ZHANG ; Sen ZHANG ; Yitong ZANG ; Kai PANG ; Xuhua HU ; Keyu REN ; Mingjuan CUI ; Shuhao LIU ; Jinhui WU ; Quan WANG ; Bo FENG ; Weidong TONG ; Yingchi YANG ; Guiying WANG ; Yun LU
Chinese Medical Journal 2025;138(21):2793-2803
BACKGROUND:
Neoadjuvant chemoradiotherapy followed by radical surgery has been a common practice for patients with locally advanced rectal cancer, but the response rate varies among patients. This study aimed to develop a ResNet-Vision Transformer based magnetic resonance imaging (MRI)-endoscopy fusion model to precisely predict treatment response and provide personalized treatment.
METHODS:
In this multicenter study, 366 eligible patients who had undergone neoadjuvant chemoradiotherapy followed by radical surgery at eight Chinese tertiary hospitals between January 2017 and June 2024 were recruited, with 2928 pretreatment colonic endoscopic images and 366 pelvic MRI images. An MRI-endoscopy fusion model was constructed based on the ResNet backbone and Transformer network using pretreatment MRI and endoscopic images. Treatment response was defined as good response or non-good response based on the tumor regression grade. The Delong test and the Hanley-McNeil test were utilized to compare prediction performance among different models and different subgroups, respectively. The predictive performance of the MRI-endoscopy fusion model was comprehensively validated in the test sets and was further compared to that of the single-modal MRI model and single-modal endoscopy model.
RESULTS:
The MRI-endoscopy fusion model demonstrated favorable prediction performance. In the internal validation set, the area under the curve (AUC) and accuracy were 0.852 (95% confidence interval [CI]: 0.744-0.940) and 0.737 (95% CI: 0.712-0.844), respectively. Moreover, the AUC and accuracy reached 0.769 (95% CI: 0.678-0.861) and 0.729 (95% CI: 0.628-0.821), respectively, in the external test set. In addition, the MRI-endoscopy fusion model outperformed the single-modal MRI model (AUC: 0.692 [95% CI: 0.609-0.783], accuracy: 0.659 [95% CI: 0.565-0.775]) and the single-modal endoscopy model (AUC: 0.720 [95% CI: 0.617-0.823], accuracy: 0.713 [95% CI: 0.612-0.809]) in the external test set.
CONCLUSION
The MRI-endoscopy fusion model based on ResNet-Vision Transformer achieved favorable performance in predicting treatment response to neoadjuvant chemoradiotherapy and holds tremendous potential for enabling personalized treatment regimens for locally advanced rectal cancer patients.
Humans
;
Rectal Neoplasms/diagnostic imaging*
;
Magnetic Resonance Imaging/methods*
;
Male
;
Female
;
Middle Aged
;
Neoadjuvant Therapy/methods*
;
Aged
;
Adult
;
Chemoradiotherapy/methods*
;
Endoscopy/methods*
;
Treatment Outcome
2.Lip and oral cancers in East Asia from 1990 to 2035: trends of disease burden and future projections.
Yitong LIU ; Ke ZHAO ; Xiaodong WANG
Journal of Southern Medical University 2025;45(7):1554-1562
OBJECTIVES:
To analyze the trends of disease burden of lip and oral cancers in East Asia from 1990 to 2021 and its future projections.
METHODS:
We used the Global Burden of Disease 2021 database to conduct a comprehensive analysis of disease burden data from China (including Taiwan Province of China), Japan, Republic of Korea, Democratic People's Republic of Korea and Mongolia. The data were stratified by age, gender and major risk factors, and a Bayesian age-period-cohort model was employed to predict the future trends.
RESULTS:
From 1990 to 2021, the burden of lip and oral cancers in East Asian countries exhibited a steady increase. Taiwan Province of China experienced the most significant increases in incidence, prevalence, mortality, and disability-adjusted life years (DALYs), while Mongolia saw a decline in both mortality and DALYs. In 2021, Taiwan Province of China reported the highest rates of lip and oral cancer incidence (27.50 per 100 000), prevalence (137.92 per 100 000), mortality (9.59 per 100 000), and DALYs (292.07 person-years per 100 000), particularly among male and elderly populations. Tobacco use and alcohol consumption significantly exacerbated the disease burden in Taiwan Province of China and Japan. Future projections indicate that the incidence and prevalence of lip and oral cancer in China (excluding Taiwan Province of China) will continue to rise, while their mortality rates are expected to decline in most regions, except for Taiwan Province of China and Democratic People's Republic of Korea.
CONCLUSIONS
By the year 2035, the disease burden of lip and oral cancers in East Asia is expected to continue to increase, especially in Taiwan Province of China. To address this challenge, it is essential to implement effective measures to control major risk factors, promote early screening, and ensure equitable distribution of healthcare resources.
Humans
;
Mouth Neoplasms/epidemiology*
;
Incidence
;
Lip Neoplasms/epidemiology*
;
Asia, Eastern/epidemiology*
;
Male
;
Disability-Adjusted Life Years
;
Prevalence
;
Female
;
Forecasting
;
Risk Factors
;
Cost of Illness
;
Middle Aged
;
Global Burden of Disease
;
Aged
;
Bayes Theorem
3.pLM4ACP: a model for predicting anticancer peptides based on machine learning and protein language models.
Yitong LIU ; Wenxin CHEN ; Juanjuan LI ; Xue CHI ; Xiang MA ; Yanqiong TANG ; Hong LI
Chinese Journal of Biotechnology 2025;41(8):3252-3261
Cancer is a serious global health problem and a major cause of human death. Conventional cancer treatments often run the risk of impairing vital organ functions. Anticancer peptides (ACPs) are considered to be one of the most promising therapeutic agents against common human cancers due to their small sizes, high specificity, and low toxicity. Since ACP recognition is highly limited to the laboratory, expensive, and time-consuming, we proposed pLM4ACP, a model for predicting ACPs based on machine learning and protein language models. In this model, the protein language model ProtT5 was used to extract the features of ACPs, and the extracted features were input into the support vector machine (SVM) classification algorithm for optimization and performance evaluation. The model showcased significantly higher accuracy than other methods, with the overall accuracy of 0.763, F1-score of 0.767, Matthews correlation coefficient of 0.527, and area under the curve of 0.827 on the independent test set. This study constructs an efficient anticancer peptide prediction model based on protein language models, further advancing the application of artificial intelligence in the biomedical field and promoting the development of precision medicine and computational biology.
Machine Learning
;
Antineoplastic Agents/chemistry*
;
Humans
;
Peptides/chemistry*
;
Support Vector Machine
;
Algorithms
;
Computational Biology/methods*
;
Neoplasms/drug therapy*
4.Temporal-spatial Generation of Astrocytes in the Developing Diencephalon.
Wentong HONG ; Pifang GONG ; Xinjie PAN ; Zhonggan REN ; Yitong LIU ; Guibo QI ; Jun-Liszt LI ; Wenzhi SUN ; Woo-Ping GE ; Chun-Li ZHANG ; Shumin DUAN ; Song QIN
Neuroscience Bulletin 2024;40(1):1-16
Astrocytes are the largest glial population in the mammalian brain. However, we have a minimal understanding of astrocyte development, especially fate specification in different regions of the brain. Through lineage tracing of the progenitors of the third ventricle (3V) wall via in-utero electroporation in the embryonic mouse brain, we show the fate specification and migration pattern of astrocytes derived from radial glia along the 3V wall. Unexpectedly, radial glia located in different regions along the 3V wall of the diencephalon produce distinct cell types: radial glia in the upper region produce astrocytes and those in the lower region produce neurons in the diencephalon. With genetic fate mapping analysis, we reveal that the first population of astrocytes appears along the zona incerta in the diencephalon. Astrogenesis occurs at an early time point in the dorsal region relative to that in the ventral region of the developing diencephalon. With transcriptomic analysis of the region-specific 3V wall and lateral ventricle (LV) wall, we identified cohorts of differentially-expressed genes in the dorsal 3V wall compared to the ventral 3V wall and LV wall that may regulate astrogenesis in the dorsal diencephalon. Together, these results demonstrate that the generation of astrocytes shows a spatiotemporal pattern in the developing mouse diencephalon.
Mice
;
Animals
;
Astrocytes
;
Neuroglia/physiology*
;
Diencephalon
;
Brain
;
Neurons
;
Mammals
5.The Influence of "Five Movements and Six Qi" Theory on Liu Wansu's Medical Theory System
Liwei LIU ; Jiang REN ; Yitong WANG ; Changhua LIU
Journal of Nanjing University of Traditional Chinese Medicine 2024;40(4):334-340
The five movements and six qi theory(hereinafter referred to as"movement qi theory")runs through Liu Wansu's med-ical theoretical system,which is mainly reflected in its influence on the theory of pathogenesis,the creation of fiery heat theory,and the influence of Liu Wansu's clinical diagnosis and treatment thoughts.The influence of movement qi theory on Liu Wansu's pathogen-esis theory includes three aspects:six qi alternation on pathogenesis,five movement hyperactivities on pathogenesis,and the root and branch of movement qi on pathogenesis.Liu's understanding of the pathogenesis of six qi and the philosophical thought of"water is good and fire is evil"mainly influenced the creation of Liu Wansu's"fire theory".The influence of movement qi theory on Liu Wansu's clinical diagnosis and treatment thoughts is mainly reflected in the classification of viscera,meridian and collateral diseases based on the theory of root and branch of middle qi,the diagnosis of pulses based on movement qi theory,the deduction of disease transmission and change using movement qi thinking,and the preparation of medicines based on qi.Liu Wansu initiated the innovation and debate of medical theory in the Jin and Yuan Dynasties,and had a profound impact on the rise of warm febrile diseases in the Ming and Qing Dynasties.
6.Efficacy of dexmedetomidine for patient-controlled sleep regulation in improving postoperative sleep disorders in patients with gastrointestinal tumors
Ziyu GU ; Yu LIU ; Yitong DING ; Zhouquan WU
Chinese Journal of Anesthesiology 2024;44(5):609-614
Objective:To evaluate the efficacy of dexmedetomidine for patient-controlled sleep regulation in improving postoperative sleep disorders in patients with gastrointestinal tumors.Methods:One hundred and fifty American Society of Anesthesiologists Physical Status classification Ⅱ or Ⅲ patients, regardless of gender, aged 40-80 yr, with a preoperative Pittsburgh Sleep Quality Index (PSQI) score of ≤7 and a PSQI score of >7 on the 1st day after surgery, undergoing elective laparoscopic resection of gastrointestinal tumors from May 2023 to December 2023 in the Affiliated Changzhou No. 2 People′s Hospital of Nanjing Medical University, were divided into 3 groups ( n=50 each) using a random number table method: normal saline group and dexmedetomidine via different routes of administration groups (DEX1 group, DEX2 group). After the routine use of 48 h postoperative analgesia, dexmedetomidine 400 μg and atropine 1 mg in 100 ml of normal saline were added to the analgesic pump in DEX1 and DEX2 groups, DEX1 group received a background infusion at a rate of 2.5 ml/h, and after an initial dose of 6 ml, the patient-controlled analgesia (PCA) pump was programmed to deliver 4 ml with a lockout interval of 10 min and background infusion at 0.5 ml/h in DEX2 group. In NS group, normal saline was added to the PCA pump, and the methed of petient-controlled administration was the same as those previously described in DEX2 group. PSQI scores were recorded at days 1, 3 and 7 and 1 month postoperatively, and visual analogue scale scores were recorded on postoperative days 1, 3 and 7. Personal Health Questionnaire Depression Scale scores were assessed and the polysomnogram was monitored on the preoperative day 1, and 15-item Quality of Recovery (QoR-15) scale scores were assessed on postoperative day 7. The duration of PACU stay, consumption of anesthetic drugs, the total pressing times of PCA within 48 h, consumption of analgesic drugs and lenth of hospital stay were recorded. Results:Compared with NS group, the sleep stage N1 ratio and arousal index were significantly decreased and the sleep stage N2 ratio and sleep efficiency were increased on postoperative days 3 and 7, PSQI scores at days 3 and 7 and 1 month after operation and VAS score at postoperative day 7 were decreased, the length of hospital stay was shortened in DEX1 and DEX2 groups, and QoR-15 scale scores were significantly increased in DEX2 group ( P<0.05). Compared with DEX1 group, the sleep stage N3 ratio was significantly increased, PSQI scores were decreased on postoperative days 3 and 7, and QoR-15 scale scores were increased in DEX2 group ( P<0.05). Multivariate logistic regression analysis showed that dexmedetomidine for patient-controlled sleep regulation was a protective factor against postoperative sleep disturbances ( P<0.05). Conclusions:For the patients with postoperative sleep disorders following surgery for gastrointestinal tumors, self-controlled administration of dexmedetomidine for 3 consecutive days after surgery improves the sleep structure, raises the subjective sleep quality and promotes the postoperative recovery.
7.Effects of low-dose esketamine on the median effective dose of ciprofol for anesthesia induction in painless gastrointestinal endoscopy
Yanhui MA ; Yiwen LIAN ; Fangyan LIU ; Ke HUANG ; Yitong JIA ; Fanqi MENG ; Jie WU ; Tianlong WANG
The Journal of Clinical Anesthesiology 2024;40(5):458-462
Objective To assess the effects of low-dose esketamine on the median effective dose(ED50)of ciprofol for anesthesia induction in painless gastrointestinal endoscopy.Methods Fifty-nine pa-tients underwent elective painless gastrointestinal endoscopy,26 males and 33 females,aged 18-64 years,BMI 18-30 kg/m2,ASA physical status Ⅰ or Ⅱ,were divided into two groups by random number table method:esketamine combined with ciprofol group(group EC,n = 29)and ciprofol group(group C,n = 30).Group EC received intravenous injection of esketamine 0.3 mg/kg and group C received the same amount of normal saline 2 minutes before administration of ciprofol.The initial anesthesia induction dose of ciprofol was 0.4 mg/kg.If a positive reaction occurs during the examination,the next patient will receive an increase dose of propofol 0.04 mg/kg,otherwise will decrease by propofol 0.04 mg/kg.The positive reaction was defined that the patient's BIS can not be decreased to 60 2 minutes after anesthesia induction,or the cough or body movement reaction occur at level 2 or above when entering the mirror.The dosage of ciprofol,recovery time,discharge time,the occurrence of intraoperative and postoperative adverse reactions were recorded.The ED50,95%effective dose(ED95)and 95%confidence interval(CI)of the two groups were calculated by Probit probability regression analysis.Results Compared with group C,the dosage of ciprofol,the incidence of hypotension and frequency of administration of vasoactive drugs during the exami-nation process in group EC were significantly reduced(P<0.05).The ED50 of ciprofol for anesthesia in-duction in painless gastrointestinal endoscopy in group EC was 0.21 mg/kg(95%CI 0.12-0.25 mg/kg)and the ED95 was 0.32 mg/kg(95%CI 0.26-0.39 mg/kg).The ED50 of ciprofol for anesthesia induction in painless gastrointestinal endoscopy in group C was 0.37 mg/kg(95%CI 0.32-0.40 mg/kg)and the ED95 was 0.48 mg/kg(95%CI 0.43-0.54 mg/kg).The ED50 and ED95 of ciprofol for anesthesia induction in painless gastrointestinal endoscopy in group EC was significantly lower than that in group C(P<0.05).There was no significant difference in other frequency of adverse events between the two groups.Conclusion Esketamine 0.3 mg/kg can reduce the ED50 of ciprofol in painless gastrointestinal endoscopy and reduce the dosage of ciprofol during the examination process,which is safe for painless gastrointestinal endoscopy with stable intraoperative circulation.
8.How do female mosquitoes determine the most suitable males for mating?
Yitong LI ; Dong LI ; Xiaofei LIU ; Ying WANG ; Tingting LIU ; Yanqiu XU ; Shengqun DENG
Chinese Journal of Schistosomiasis Control 2024;36(3):321-328
More than 80% of the world’s populations are at risk of vector-borne diseases, with mosquito-borne diseases as a significant global public health problem. Mosquito populations control is critical to interrupting the transmission of mosquito-borne diseases. This review summarizes the physical attributes, smell, vision, touch, and hearing of mosquitoes to unravel the preferences of female mosquitoes, and describes the mechanisms underlying the best male mating by female mosquitoes, so as to provide new insights into management of mosquito-borne diseases.
9.Advances in the application of 5×FAD transgenic mice in Alzheimer's disease research
Yiduan LIU ; Yixiao LIU ; Xinyuan HAN ; Yitong XIAO ; Tianyuan YE
Acta Laboratorium Animalis Scientia Sinica 2024;32(2):260-274
Transgenic 5 × FAD mice are APP/PS1 transgenic mice carrying five familial Alzheimer's disease(AD)gene mutations.Beta-amyloid precursor protein(amyloid precursor protein,APP)expression is related to the K670N/M671L(Swedish),1716V(Florida),and V7171(London)mutations,and presenilin 1(PSI)is affected by the M146L and L286V mutations.5 × FAD mice express high levels of β-amyloid in the brain at 1.5 months old,and neuritic plaques began to appear at 2 months old.The pathological phenotypes of 5 × FAD mice include amyloid plaque aggregation,neuronal loss,gliosis,and memory dysfunction,while their biological characteristics include changes in the formation of brain β-amyloid plaques,hyperphosphorylation of Tau protein,synaptic dysfunction,neuroinflammatory response,mitochondrial dysfunction,blood-brain barrier injury,neuronal injury,endoplasmic reticulum stress,and eye lesions.As a classic animal model of AD,5 × FAD transgenic mice can simulate the neuropathological process and behavioral manifestations of late-stage AD in humans,and these mice are thus widely used in research into the pathogenesis of AD and the development of new drugs.In this review,we summarize the model construction,biological background,and biological characteristics of 5 x FAD transgenic mice,and the development and application of drugs for the prevention and treatment of AD,to provide references for the application of 5 x FAD transgenic transgenic mice in AD research.
10.Advances in applications of neuroregulatory techniques in research on brain sciences
Mengnan LIU ; Xiaoyu TIAN ; Yitong LI ; Ning WU ; Jin LI ; Hong LI
Chinese Journal of Pharmacology and Toxicology 2024;38(2):128-136
Drugs and physical stimulation,including light,electricity,and magnetic fields,can be used to influence how neurons operate,among which chemogenetic and optogenetic technologies are most widely used.In recent years,magnetogenetic technology has also been developed that can acti-vate neurons in magnetic fields through magnetic sensitive actuators,leading to non-invasive and instanta-neous activation of specific brain regions.This article reviews the evolution of and problems with chemoge-netic and optogenetic techniques commonly used in brain science research.It also outlines the latest progress in magnetogenetic technologies,which are not full-fledged yet,as well as the role of transcra-nial electrical stimulation,transcranial magnetic stimulation,deep brain stimulation and transcranial ultra-sound stimulation technology in the functional regulation of brain diseases.Constant adjustment and improvement can make it possible for these technologies to be used more widely for the study of brain sciences and the treatment of brain diseases.

Result Analysis
Print
Save
E-mail