Visual Attention Prediction(VAP)is widely applied in GIS research,such as navigation task identification and driver assistance systems.Previous studies commonly took color information to detect the visual saliency of ...Visual Attention Prediction(VAP)is widely applied in GIS research,such as navigation task identification and driver assistance systems.Previous studies commonly took color information to detect the visual saliency of natural scene images.However,these studies rarely considered adaptively feature integration to different geospatial scenes in specific tasks.To better predict visual attention while driving tasks,in this paper,we firstly propose an Adaptive Feature Integration Fully Convolutional Network(AdaFI-FCN)using Scene-Adaptive Weights(SAW)to integrate RGB-D,motion and semantic features.The quantitative comparison results on the DR(eye)VE dataset show that the proposed framework achieved the best accuracy and robustness performance compared with state-of-the-art models(AUC-Judd=0.971,CC=0.767,KL=1.046,SIM=0.579).In addition,the experimental results of the ablation study demonstrated the positive effect of the SAW method on the prediction robustness in response to scene changes.The proposed model has the potential to benefit adaptive VAP research in universal geospatial scenes,such as AR-aided navigation,indoor navigation,and street-view image reading.展开更多
In this paper, we present a novel and efficient scheme for extracting, indexing and retrieving color images. Our motivation was to reduce the space overhead of partition-based approaches taking advantage of the fact t...In this paper, we present a novel and efficient scheme for extracting, indexing and retrieving color images. Our motivation was to reduce the space overhead of partition-based approaches taking advantage of the fact that only a relatively low number of distinct values of a particular visual feature is present in most images. To extract color feature and build indices into our image database we take into consideration factors such as human color perception and perceptual range, and the image is partitioned into a set of regions by using a simple classifying scheme. The compact color feature vector and the spatial color histogram, which are extracted from the seqmented image region, are used for representing the color and spatial information in the image. We have also developed the region-based distance measures to compare the similarity of two images. Extensive tests on a large image collection were conducted to demonstrate the effectiveness of the proposed approach.展开更多
The anomaly detection of electromagnetic environment situation(EMES) has essential reference value for electromagnetic equipment behavior cognition and battlefield threat assessment.In this paper,we proposed a deep le...The anomaly detection of electromagnetic environment situation(EMES) has essential reference value for electromagnetic equipment behavior cognition and battlefield threat assessment.In this paper,we proposed a deep learning-based method for detecting anomalies in EMES to address the problem of relatively low efficiency of electromagnetic environment situation anomaly detection(EMES-AD).Firstly,the convolutional kernel extracts the static features of different regions of the EMES.Secondly,the dynamic features of the region are obtained by using a recurrent neural network(LSTM).Thirdly,the Spatio-temporal features of the region are recovered by using a de-convolutional network and then fused to predict the EMES.The structural similarity algorithm(SSIM) is used to determine whether it is anomalous.We developed the detection framework,de-signed the network parameters,simulated the data sets containing different anomalous types of EMES,and carried out the detection experiments.The experimental results show that the proposed method is effective.展开更多
Bioinspired energy‐autonomous interactive electronics are prevalent.However,self‐powered artificial skins are often challenging to be combined with excellent mechanical properties,optical transparency,autonomous att...Bioinspired energy‐autonomous interactive electronics are prevalent.However,self‐powered artificial skins are often challenging to be combined with excellent mechanical properties,optical transparency,autonomous attachability,and biocompatibility.Herein,a robust ecological polyionic skin(polyionic eco‐skin)based on triboelectric mechanism consisting of ethyl cellulose/waterborne polyurethane/Cu nanoparticles(EWC)green electroactive sensitive material and polyethylene oxide/waterborne polyurethane/phytic acid(PWP)polyionic current collector is proposed.The polyionic eco‐skin features sufficient stretchability(90%)and low Young's modulus(0.8MPa)close to that of human soft tissue,high transparency(>84%of transmission)in the visible light range,and broad static/dynamic adhesiveness,which endows it with strong adaptive implementation capacity in flexible curved electronics.More importantly,the self‐powered polyionic eco‐skin exhibits enhanced force‐electric conversion performance by coordinating the effect of nanoparticlepolymer interfacial polarization and porous structure of sensitive material.Integrating multiple characteristics enables the polyionic ecoskin to effectively convert biomechanical energy into electrical energy,supporting self‐powered functionality for itself and related circuits.Moreover,the eco‐skin can be utilized to construct an interactive system and realize the remote noncontact manipulation of targets.The polyionic eco‐skin holds tremendous application potential in self‐powered security systems,human-machine interaction interfaces,and bionic robots,which is expected to inject new vitality into a human-cyber-physical intelligence integration.展开更多
Introduction:Human influenza A/H3N2 imposes a substantial global disease burden.Beyond hemagglutinin(HA),neuraminidase(NA)also plays a critical role in the antigenic evolution of influenza viruses.However,a comprehens...Introduction:Human influenza A/H3N2 imposes a substantial global disease burden.Beyond hemagglutinin(HA),neuraminidase(NA)also plays a critical role in the antigenic evolution of influenza viruses.However,a comprehensive understanding of NA antigenic evolution remains lacking.Methods:NA inhibition(NAI)data were collected and structural epitopes for A/H3N2 NA were identified.A machine learning model was developed to accurately predict antigenic relationships by integrating four feature groups:epitopes,physicochemical properties,N-glycosylation,and catalytic sites.An antigenic correlation network(ACNet)was constructed and antigenic clusters were identified using the Markov clustering algorithm.Results:The best random forest model(PREDEC-N2)achieved an accuracy of 0.904 in crossvalidation and 0.867 in independent testing.Eight main antigenic clusters were identified on the ACNet.Spatiotemporal analysis revealed the continuous replacement and rapid global spread of new antigenic clusters for human influenza A/H3N2 NA.Conclusions:This study developed a timely and accurate computational model to map the antigenic landscape of A/H3N2 NA,revealing both its relative antigenic conservation and continuous evolution.These insights provide valuable guidance for improved antigenic surveillance,vaccine recommendations,and prevention and control strategies for human influenza viruses.展开更多
In machine translation(MT) practice,there is an urgent need for constructing a set of Chinese-to-English aspect transferring rules to define the transferring conditions.The integrated feature set was used to generaliz...In machine translation(MT) practice,there is an urgent need for constructing a set of Chinese-to-English aspect transferring rules to define the transferring conditions.The integrated feature set was used to generalize and justify the Chinese-to-English transferring rule of the 'ZHE' aspect(ZHE Rule).A ZHE classification model was built in this study.The impacts of each set of temporal,lexical aspectual,and syntactic features,and their integrated impacts,on the accuracy of the ZHE Rule were tested.Over 600 misclassified corpus sentences were manually examined.A 10-fold cross-validation was used with a decision tree algorithm.The main results are:(1) The ZHE Rule was generalized and justified to have a higher accuracy under the two metrics:the precision rate and the areas under the receiver operating characteristic curve(AUC).(2) The temporal,lexical aspectual,and syntactic feature sets have an integrated contribution to the accuracy of the ZHE Rule.The syntactic and temporal features have an impact on ZHE aspect derivations,while the lexical aspectual features are not predictive of ZHE aspect derivation.(3) While associated with active verbs,the ZHE aspect can denote a perfective situation.This study suggests that the temporal and syntactic features are the predictive ZHE aspect classification features and that the ZHE Rule with an overall precision rate of 80.1% is accurate enough to be further explored in MT practice.The machine learning method,decision tree,can be applied to the automatic aspect transferring in MT research and aspectual interpretations in linguistic research.展开更多
Aiming at the problem that honeycomb lung lesions are difficult to accurately segment due to diverse morphology and complex distribution,a network with parallel two-branch structure is proposed.In the encoder,the Pyra...Aiming at the problem that honeycomb lung lesions are difficult to accurately segment due to diverse morphology and complex distribution,a network with parallel two-branch structure is proposed.In the encoder,the Pyramid Pooling Transformer(P2T)backbone is used as the Transformer branch to obtain the global features of the lesions,the convolutional branch is used to extract the lesions’local feature information,and the feature fusion module is designed to effectively fuse the features in the dual branches;subsequently,in the decoder,the channel prior convolutional attention is used to enhance the localization ability of the model to the lesion region.To resolve the problem of model accuracy degradation caused by the class imbalance of the dataset,an adaptive weighted hybrid loss function is designed for model training.Finally,extensive experimental results show that the method in this paper performs well on the Honeycomb Lung Dataset,with Intersection over Union(IoU),mean Intersection over Union(mIoU),Dice coefficient,and Precision(Pre)of 0.8750,0.9363,0.9298,and 0.9012,respectively,which are better than other methods.In addition,its IoU and Dice coefficient of 0.7941 and 0.8875 on the Covid dataset further prove its excellent performance.展开更多
This Special Section of the INTEGRATED CIRCUITS AND SYSTEMS(ICAS)features selected papers from the 2023 IEEE 15th International Conference ON ASIC(ASICON),held in Nanjing,Jiangsu,China,from October 24th to 27th,2023.A...This Special Section of the INTEGRATED CIRCUITS AND SYSTEMS(ICAS)features selected papers from the 2023 IEEE 15th International Conference ON ASIC(ASICON),held in Nanjing,Jiangsu,China,from October 24th to 27th,2023.ASICON is an IEEE conference in the field of integrated circuits(ICs)in China,designed to provide an international forum for IC designers,ASIC users,system integrators,IC manufacturers,process and device engineers,and CAD/CAE tool developers to showcase their latest advancements,development and research findings.展开更多
Spikes detection and sorting play an important role in study of neural information coding.Spikes were generally obtained by threshold detection after filtered in traditional detection,which failed to suppress the rand...Spikes detection and sorting play an important role in study of neural information coding.Spikes were generally obtained by threshold detection after filtered in traditional detection,which failed to suppress the random pulse interference(RPI),In this paper,a novel algorithm was provided to suppress RPI using integrated feature.The raw neural signals from the primary visual cortex in rats were detected with microelectrode array.After the feature differences between spikes and RPls were compared,the features which include waveform and non-waveform features were extracted respectively,and then the integrated feature was established based on Fisher's discrimi nant ratio to separate between spikes and RPls.The test results of simulation and experiment show that the separability capability of the integrated feature is nearly two times greater than the individual feature,the average correct recognition rate of spikes and RPls is over 93%,and the detection rate of spike is effectively improved.展开更多
Accurate prognosis prediction is essential for guiding cancer treatment and improving patient outcomes.While recent studies have demonstrated the potential of histopathological images in survival analysis,existing mod...Accurate prognosis prediction is essential for guiding cancer treatment and improving patient outcomes.While recent studies have demonstrated the potential of histopathological images in survival analysis,existing models are typically developed in a cancerspecific manner,lack extensive external validation,and often rely on molecular data that are not routinely available in clinical practice.To address these limitations,we present PROGPATH,a unified model capable of integrating histopathological image features with routinely collected clinical variables to achieve pancancer prognosis prediction.PROGPATH employs a weakly supervised deep learning architecture built upon the foundation model for image encoding.Morphological features are aggregated through an attention-guided multiple instance learning module and fused with clinical information via a cross-attention transformer.A router-based classification strategy further refines the prediction performance.PROGPATH was trained on 7999 whole-slide images(WSIs)from 6,670 patients across 15 cancer types,and extensively validated on 17 external cohorts with a total of 7374 WSIs from 4441 patients,covering 12 cancer types from 8 consortia and institutions across three continents.PROGPATH achieved consistently superior performance compared with state-of-the-art multimodal prognosis prediction models.It demonstrated strong generalizability across cancer types and robustness in stratified subgroups,including early-and advancedstage patients,treatment cohorts(radiotherapy and pharmaceutical therapy),and biomarker-defined subsets.We further provide model interpretability by identifying pathological patterns critical to PROGPATH’s risk predictions,such as the degree of cell differentiation and extent of necrosis.Together,these results highlight the potential of PROGPATH to support pancancer outcome prediction and inform personalized cancer management strategies.展开更多
基金supported by the National Natural Science Foundation of China(NSFC)under Grant No.42230103the State Key Laboratory of Geographic Information Engineering and the Key Laboratory of Surveying and Mapping Science and Geospatial Information Technology of the Ministry of Natural Resources Jointly Funded Project under Grant No.2021-04-03.
文摘Visual Attention Prediction(VAP)is widely applied in GIS research,such as navigation task identification and driver assistance systems.Previous studies commonly took color information to detect the visual saliency of natural scene images.However,these studies rarely considered adaptively feature integration to different geospatial scenes in specific tasks.To better predict visual attention while driving tasks,in this paper,we firstly propose an Adaptive Feature Integration Fully Convolutional Network(AdaFI-FCN)using Scene-Adaptive Weights(SAW)to integrate RGB-D,motion and semantic features.The quantitative comparison results on the DR(eye)VE dataset show that the proposed framework achieved the best accuracy and robustness performance compared with state-of-the-art models(AUC-Judd=0.971,CC=0.767,KL=1.046,SIM=0.579).In addition,the experimental results of the ablation study demonstrated the positive effect of the SAW method on the prediction robustness in response to scene changes.The proposed model has the potential to benefit adaptive VAP research in universal geospatial scenes,such as AR-aided navigation,indoor navigation,and street-view image reading.
文摘In this paper, we present a novel and efficient scheme for extracting, indexing and retrieving color images. Our motivation was to reduce the space overhead of partition-based approaches taking advantage of the fact that only a relatively low number of distinct values of a particular visual feature is present in most images. To extract color feature and build indices into our image database we take into consideration factors such as human color perception and perceptual range, and the image is partitioned into a set of regions by using a simple classifying scheme. The compact color feature vector and the spatial color histogram, which are extracted from the seqmented image region, are used for representing the color and spatial information in the image. We have also developed the region-based distance measures to compare the similarity of two images. Extensive tests on a large image collection were conducted to demonstrate the effectiveness of the proposed approach.
基金funded by the National Natural Science Foundation of China, grant number 11975307the National Defense Science and Technology Innovation Special Zone Project, grant number 19-H863-01-ZT-003-003-12。
文摘The anomaly detection of electromagnetic environment situation(EMES) has essential reference value for electromagnetic equipment behavior cognition and battlefield threat assessment.In this paper,we proposed a deep learning-based method for detecting anomalies in EMES to address the problem of relatively low efficiency of electromagnetic environment situation anomaly detection(EMES-AD).Firstly,the convolutional kernel extracts the static features of different regions of the EMES.Secondly,the dynamic features of the region are obtained by using a recurrent neural network(LSTM).Thirdly,the Spatio-temporal features of the region are recovered by using a de-convolutional network and then fused to predict the EMES.The structural similarity algorithm(SSIM) is used to determine whether it is anomalous.We developed the detection framework,de-signed the network parameters,simulated the data sets containing different anomalous types of EMES,and carried out the detection experiments.The experimental results show that the proposed method is effective.
基金supported by the National Natural Science Foundation of China(52403284,52473270 and T2422028)the Natural Science Foundation of Jiangsu Province(BK20240481)+3 种基金the fellowship of China National Postdoctoral Program for Innovative Talents(BX20240408)the National Key R&D Program of China(2022YFA1203304)the Suzhou Institute of Nano‐Tech and Nano‐Bionics,Chinese Academy of Sciences(start‐up grant E1552102)Jiangsu Funding Program for Excellent Postdoctoral Talent.
文摘Bioinspired energy‐autonomous interactive electronics are prevalent.However,self‐powered artificial skins are often challenging to be combined with excellent mechanical properties,optical transparency,autonomous attachability,and biocompatibility.Herein,a robust ecological polyionic skin(polyionic eco‐skin)based on triboelectric mechanism consisting of ethyl cellulose/waterborne polyurethane/Cu nanoparticles(EWC)green electroactive sensitive material and polyethylene oxide/waterborne polyurethane/phytic acid(PWP)polyionic current collector is proposed.The polyionic eco‐skin features sufficient stretchability(90%)and low Young's modulus(0.8MPa)close to that of human soft tissue,high transparency(>84%of transmission)in the visible light range,and broad static/dynamic adhesiveness,which endows it with strong adaptive implementation capacity in flexible curved electronics.More importantly,the self‐powered polyionic eco‐skin exhibits enhanced force‐electric conversion performance by coordinating the effect of nanoparticlepolymer interfacial polarization and porous structure of sensitive material.Integrating multiple characteristics enables the polyionic ecoskin to effectively convert biomechanical energy into electrical energy,supporting self‐powered functionality for itself and related circuits.Moreover,the eco‐skin can be utilized to construct an interactive system and realize the remote noncontact manipulation of targets.The polyionic eco‐skin holds tremendous application potential in self‐powered security systems,human-machine interaction interfaces,and bionic robots,which is expected to inject new vitality into a human-cyber-physical intelligence integration.
基金Supported by the National Key Research and Development Program under grant 2022YFC2303800the Major Program of Guangzhou National Laboratory under grant GZNL2024A01002+2 种基金the National Natural Science Foundation of China under grant 81961128002the Science and Technology Planning Project of Guangdong Province,China under grant 2021B1212040017the Shenzhen Science and Technology Program under grant ZDSYS20230626091203007.
文摘Introduction:Human influenza A/H3N2 imposes a substantial global disease burden.Beyond hemagglutinin(HA),neuraminidase(NA)also plays a critical role in the antigenic evolution of influenza viruses.However,a comprehensive understanding of NA antigenic evolution remains lacking.Methods:NA inhibition(NAI)data were collected and structural epitopes for A/H3N2 NA were identified.A machine learning model was developed to accurately predict antigenic relationships by integrating four feature groups:epitopes,physicochemical properties,N-glycosylation,and catalytic sites.An antigenic correlation network(ACNet)was constructed and antigenic clusters were identified using the Markov clustering algorithm.Results:The best random forest model(PREDEC-N2)achieved an accuracy of 0.904 in crossvalidation and 0.867 in independent testing.Eight main antigenic clusters were identified on the ACNet.Spatiotemporal analysis revealed the continuous replacement and rapid global spread of new antigenic clusters for human influenza A/H3N2 NA.Conclusions:This study developed a timely and accurate computational model to map the antigenic landscape of A/H3N2 NA,revealing both its relative antigenic conservation and continuous evolution.These insights provide valuable guidance for improved antigenic surveillance,vaccine recommendations,and prevention and control strategies for human influenza viruses.
基金supported by the National Social Science Foundation of China(No.08BYY001)the Worldwide Universities Network 2009 Research Mobility Programme
文摘In machine translation(MT) practice,there is an urgent need for constructing a set of Chinese-to-English aspect transferring rules to define the transferring conditions.The integrated feature set was used to generalize and justify the Chinese-to-English transferring rule of the 'ZHE' aspect(ZHE Rule).A ZHE classification model was built in this study.The impacts of each set of temporal,lexical aspectual,and syntactic features,and their integrated impacts,on the accuracy of the ZHE Rule were tested.Over 600 misclassified corpus sentences were manually examined.A 10-fold cross-validation was used with a decision tree algorithm.The main results are:(1) The ZHE Rule was generalized and justified to have a higher accuracy under the two metrics:the precision rate and the areas under the receiver operating characteristic curve(AUC).(2) The temporal,lexical aspectual,and syntactic feature sets have an integrated contribution to the accuracy of the ZHE Rule.The syntactic and temporal features have an impact on ZHE aspect derivations,while the lexical aspectual features are not predictive of ZHE aspect derivation.(3) While associated with active verbs,the ZHE aspect can denote a perfective situation.This study suggests that the temporal and syntactic features are the predictive ZHE aspect classification features and that the ZHE Rule with an overall precision rate of 80.1% is accurate enough to be further explored in MT practice.The machine learning method,decision tree,can be applied to the automatic aspect transferring in MT research and aspectual interpretations in linguistic research.
基金supported by the Central Leading Local Science and Technology Development Fund(Nos.YDZJSX2021C004 and YDZJSX20231C004)the Natural Science Foundation of Shanxi Province(No.20210302124554).
文摘Aiming at the problem that honeycomb lung lesions are difficult to accurately segment due to diverse morphology and complex distribution,a network with parallel two-branch structure is proposed.In the encoder,the Pyramid Pooling Transformer(P2T)backbone is used as the Transformer branch to obtain the global features of the lesions,the convolutional branch is used to extract the lesions’local feature information,and the feature fusion module is designed to effectively fuse the features in the dual branches;subsequently,in the decoder,the channel prior convolutional attention is used to enhance the localization ability of the model to the lesion region.To resolve the problem of model accuracy degradation caused by the class imbalance of the dataset,an adaptive weighted hybrid loss function is designed for model training.Finally,extensive experimental results show that the method in this paper performs well on the Honeycomb Lung Dataset,with Intersection over Union(IoU),mean Intersection over Union(mIoU),Dice coefficient,and Precision(Pre)of 0.8750,0.9363,0.9298,and 0.9012,respectively,which are better than other methods.In addition,its IoU and Dice coefficient of 0.7941 and 0.8875 on the Covid dataset further prove its excellent performance.
基金FRANCOIS RIVET,(Senior Member,IEEE)received the M.Sc.degree in electronics from the Bordeaux Institute of Technology(Bordeaux INP),Gradignan,France,in 2005,and the Ph.D.degree in electronics from Universite de Bordeaux,Bordeaux,France,in 2009.He is currently an Associate Professor with the Electrical Engineering Department,Bordeaux Institute of Technology,and IMS Laboratory,Bordeaux.Since 2014,he has been leading the research team Circuits and Systems.He has contributed to the design of disruptive communication circuits developing a design by mathematics methodology.He has authored or coauthored 150 technical papers and holds 19 patents.His research interests include the design of integrated circuits and systems for wireless communications.francois.rivet@ims-bordeaux.frLIANG QI,(Senior Member,IEEE)received the Ph.D.degree in electrical and computer engineering from the University of Macao,Macao,in 2019.He was with Shanghai Hisilicon,where he conducted the project of multi-band(2G-5G)RX ADC.He was a Visiting Scholar with Ulm University,Ulm,Germany,during the Ph.D.studies.He is currently an Associate Professor with the Department of Micro and Nano Electronics,Shanghai Jiao Tong University.His research interests include high-performance data converters and analog mixed-signal integrated circuits.Dr.Qi was an Associate Editor for IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS—II:EXPRESS BRIEFS.He is/has been a TPC Member for IEEE APCCAS,ICSICT,ICTA,and ASICON.He was the recipient of Macao Scientific and Technology Research and Development for Postgraduate Award in 2016 and Outstanding Young Scholar Paper Award in IEEE ASICON 2021.qi.liang@sjtu.edu.cn。
文摘This Special Section of the INTEGRATED CIRCUITS AND SYSTEMS(ICAS)features selected papers from the 2023 IEEE 15th International Conference ON ASIC(ASICON),held in Nanjing,Jiangsu,China,from October 24th to 27th,2023.ASICON is an IEEE conference in the field of integrated circuits(ICs)in China,designed to provide an international forum for IC designers,ASIC users,system integrators,IC manufacturers,process and device engineers,and CAD/CAE tool developers to showcase their latest advancements,development and research findings.
基金National Natural Science Foundation of China,grant number:60841004,60971110
文摘Spikes detection and sorting play an important role in study of neural information coding.Spikes were generally obtained by threshold detection after filtered in traditional detection,which failed to suppress the random pulse interference(RPI),In this paper,a novel algorithm was provided to suppress RPI using integrated feature.The raw neural signals from the primary visual cortex in rats were detected with microelectrode array.After the feature differences between spikes and RPls were compared,the features which include waveform and non-waveform features were extracted respectively,and then the integrated feature was established based on Fisher's discrimi nant ratio to separate between spikes and RPls.The test results of simulation and experiment show that the separability capability of the integrated feature is nearly two times greater than the individual feature,the average correct recognition rate of spikes and RPls is over 93%,and the detection rate of spike is effectively improved.
基金supported in part by the National Cancer Institute under award numbers R01CA268287A1,U01CA269181,R01CA26820701A1,R01CA249992-01A1,R01CA202752-01A1,R01CA208236-01A1,R01CA216579-01A1,R01CA220581-01A1,R01CA257612-01A1,1U01CA239055-01,1U01CA248226-01,1U54CA254566-01National Heart,Lung and Blood Institute 1R01HL15127701A1,R01HL15807101A1+8 种基金National Institute of Biomedical Imaging and Bioengineering 1R43EB028736-01VA Merit Review Award IBX004121A from the United States Department of Veterans Affairs Biomedical Laboratory Research and Development Service the Office of the Assistant Secretary of Defense for Health Affairs,through the Breast Cancer Research Program(W81XWH-19-1-0668)the Prostate Cancer Research Program(W81XWH-20-1-0851)the Lung Cancer Research Program(W81XWH-18-1-0440,W81XWH-20-1-0595)the Peer Reviewed Cancer Research Program(W81XWH-18-1-0404,W81XWH-21-1-0345,W81XWH-211-0160)the Kidney Precision Medicine Project(KPMP)Glue Grant and sponsored research agreements from Bristol Myers-Squibb,Boehringer-Ingelheim,Eli-Lilly and Astrazenecasupported in part by the National Natural Science Foundation of China general program(No.61571314)the Sichuan University-Yibin City Strategic Cooperation Special Fund(No.2020CDYB-27)Support Program of Sichuan Science and Technology Department(No.2023YFS0327-LH).
文摘Accurate prognosis prediction is essential for guiding cancer treatment and improving patient outcomes.While recent studies have demonstrated the potential of histopathological images in survival analysis,existing models are typically developed in a cancerspecific manner,lack extensive external validation,and often rely on molecular data that are not routinely available in clinical practice.To address these limitations,we present PROGPATH,a unified model capable of integrating histopathological image features with routinely collected clinical variables to achieve pancancer prognosis prediction.PROGPATH employs a weakly supervised deep learning architecture built upon the foundation model for image encoding.Morphological features are aggregated through an attention-guided multiple instance learning module and fused with clinical information via a cross-attention transformer.A router-based classification strategy further refines the prediction performance.PROGPATH was trained on 7999 whole-slide images(WSIs)from 6,670 patients across 15 cancer types,and extensively validated on 17 external cohorts with a total of 7374 WSIs from 4441 patients,covering 12 cancer types from 8 consortia and institutions across three continents.PROGPATH achieved consistently superior performance compared with state-of-the-art multimodal prognosis prediction models.It demonstrated strong generalizability across cancer types and robustness in stratified subgroups,including early-and advancedstage patients,treatment cohorts(radiotherapy and pharmaceutical therapy),and biomarker-defined subsets.We further provide model interpretability by identifying pathological patterns critical to PROGPATH’s risk predictions,such as the degree of cell differentiation and extent of necrosis.Together,these results highlight the potential of PROGPATH to support pancancer outcome prediction and inform personalized cancer management strategies.
