{"title":"Interpretable Predictive Models for Healthcare via Rational Multi-Layer Perceptrons","authors":"Thiti Suttaket, Stanley Kok","doi":"10.1145/3671150","DOIUrl":"https://doi.org/10.1145/3671150","url":null,"abstract":"The healthcare sector has recently experienced an unprecedented surge in digital data accumulation, especially in the form of electronic health records (EHRs). These records constitute a precious resource that Information Systems (IS) researchers could utilize for various clinical applications, such as morbidity prediction and risk stratification. Recently, deep learning has demonstrated state-of-the-art empirical results in terms of predictive performance on EHRs. However, the blackbox nature of deep learning models prevents both clinicians and patients from trusting the models, especially with regards to life-critical decision making. To mitigate this, attention mechanisms are normally employed to improve the transparency of deep learning models. However, these mechanisms can only highlight important inputs without sufficient clarity on how they correlate with each other and still confuse end-users. To address this drawback, we pioneer a novel model called Rational Multi-Layer Perceptrons (RMLP) that is constructed from weighted finite state automata. RMLP is able to provide better interpretability by coherently linking together relevant inputs at different timesteps into distinct sequences. RMLP can be shown to be a generalization of a multi-layer perceptron (that only works on static data) to sequential, dynamic data. With its theoretical roots in rational series, RMLP’s ability to process longitudinal time-series data and extract interpretable patterns sets it apart. Using real-world EHRs, we have substantiated the effectiveness of our RMLP model through empirical comparisons on six clinical tasks, all of which demonstrate its considerable efficacy.","PeriodicalId":45274,"journal":{"name":"ACM Transactions on Management Information Systems","volume":null,"pages":null},"PeriodicalIF":2.5,"publicationDate":"2024-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"141378371","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Jessica Qiuhua Sheng, Da Xu, Paul Jen-Hwa Hu, Liang Li, Ting-Shuo Huang
{"title":"Mining Multimorbidity Trajectories and Co-Medication Effects from Patient Data to Predict Post–Hip Fracture Outcomes","authors":"Jessica Qiuhua Sheng, Da Xu, Paul Jen-Hwa Hu, Liang Li, Ting-Shuo Huang","doi":"10.1145/3665250","DOIUrl":"https://doi.org/10.1145/3665250","url":null,"abstract":"Hip fractures have profound impacts on patients’ conditions and quality of life, even when they receive therapeutic treatments. Many patients face the risk of poor prognosis, physical impairment, and even mortality, especially older patients. Accurate patient outcome estimates after an initial fracture are critical to physicians’ decision-making and patient management. Effective predictions might benefit from analyses of patients’ multimorbidity trajectories and medication usages. If adequately modeled and analyzed, they could help identify patients at higher risk of recurrent fractures or mortality. Most analytics methods overlook the onset, co-occurrence, and temporal sequence of distinct chronic diseases in the trajectory, and they also seldom consider the combined effects of different medications. To support effective predictions, we develop a novel deep learning–based method that uses a cross-attention mechanism to model patient progression by obtaining “contextual information” from multimorbidity trajectories. This method also incorporates a nested self-attention network that captures the combined effects of distinct medications by learning the interactions among medications and how dosages might influence post-fracture outcomes. A real-world patient data set is used to evaluate the proposed method, relative to six benchmark methods. The comparative results indicate that our method consistently outperforms all the benchmarks in precision, recall, F-measures, and area under the curve. The proposed method is generalizable and can be implemented as a decision support system to identify patients at greater risk of recurrent hip fractures or mortality, which should help clinical decision-making and patient management.","PeriodicalId":45274,"journal":{"name":"ACM Transactions on Management Information Systems","volume":null,"pages":null},"PeriodicalIF":2.5,"publicationDate":"2024-05-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140964914","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"ShennongMGS: An LLM-based Chinese Medication Guidance System","authors":"Yutao Dou, Yuwei Huang, Xiongjun Zhao, Haitao Zou, Jiandong Shang, Ying Lu, Xiaolin Yang, Jian Xiao, Shaoliang Peng","doi":"10.1145/3658451","DOIUrl":"https://doi.org/10.1145/3658451","url":null,"abstract":"The rapidly evolving field of Large Language Models (LLMs) holds immense promise for healthcare, particularly in medication guidance and adverse drug reaction prediction. Despite their potential, existing LLMs face challenges in dealing with complex polypharmacy scenarios and often grapple with data lag issues. To address these limitations, we introduce an LLM-based Chinese medication guidance system, called ShennongMGS, specifically tailored for robust medication guidance and adverse drug reaction predictions. Our system transforms multi-source heterogeneous medication information into a knowledge graph and employs a two-stage training strategy to construct a specialised LLM (ShennongGPT). This method enables the simulation of professional pharmacists’ decision-making processes and incorporates the capability for knowledge self-updating, thereby significantly enhancing drug safety and the overall quality of medical services. Rigorously evaluated by medical professionals and artificial intelligence experts, our method demonstrates superiority, outperforming existing general and specialised LLMs in performance.","PeriodicalId":45274,"journal":{"name":"ACM Transactions on Management Information Systems","volume":null,"pages":null},"PeriodicalIF":2.5,"publicationDate":"2024-04-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140691422","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Youxi Wu, Zhen Wang, Yan Li, Ying Guo, He Jiang, Xingquan Zhu, Xindong Wu
{"title":"Co-occurrence order-preserving pattern mining with keypoint alignment for time series","authors":"Youxi Wu, Zhen Wang, Yan Li, Ying Guo, He Jiang, Xingquan Zhu, Xindong Wu","doi":"10.1145/3658450","DOIUrl":"https://doi.org/10.1145/3658450","url":null,"abstract":"Recently, order-preserving pattern (OPP) mining has been proposed to discover some patterns, which can be seen as trend changes in time series. Although existing OPP mining algorithms have achieved satisfactory performance, they discover all frequent patterns. However, in some cases, users focus on a particular trend and its associated trends. To efficiently discover trend information related to a specific prefix pattern, this paper addresses the issue of co-occurrence OPP mining (COP) and proposes an algorithm named COP-Miner to discover COPs from historical time series. COP-Miner consists of three parts: extracting keypoints, preparation stage, and iteratively calculating supports and mining frequent COPs. Extracting keypoints is used to obtain local extreme points of patterns and time series. The preparation stage is designed to prepare for the first round of mining, which contains four steps: obtaining the suffix OPP of the keypoint sub-time series, calculating the occurrences of the suffix OPP, verifying the occurrences of the keypoint sub-time series, and calculating the occurrences of all fusion patterns of the keypoint sub-time series. To further improve the efficiency of support calculation, we propose a support calculation method with an ending strategy that uses the occurrences of prefix and suffix patterns to calculate the occurrences of superpatterns. Experimental results indicate that COP-Miner outperforms the other competing algorithms in running time and scalability. Moreover, COPs with keypoint alignment yield better prediction performance.","PeriodicalId":45274,"journal":{"name":"ACM Transactions on Management Information Systems","volume":null,"pages":null},"PeriodicalIF":2.5,"publicationDate":"2024-04-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140708331","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Estimating Future Financial Development of Urban Areas for Deploying Bank Branches: A Local-Regional Interpretable Model","authors":"Pei-Xuan Li, Yu-En Chang, Ming-Chun Wei, Hsun-Ping Hsieh","doi":"10.1145/3656479","DOIUrl":"https://doi.org/10.1145/3656479","url":null,"abstract":"\u0000 Financial forecasting is an important task for urban development. In this paper, we propose a novel deep learning framework to predict the future financial potential of urban spaces. To be more precise, our target is to infer the number of financial institutions in the future for any arbitrary location with environmental and geographical data. We propose a novel local-regional model, the\u0000 L\u0000 ocal-Regional\u0000 I\u0000 nterpretable\u0000 M\u0000 ulti-\u0000 A\u0000 ttention model (LIMA model), that considers multiple aspects of a location - the place itself and its surroundings. Besides, our model offers three kinds of interpretability, providing a superior way for decision makers to understand how the model determines the prediction: critical rules learned from the tree-based module, surrounding locations that are high-correlated with the prediction, and critical regional features. Our module not only takes advantage of a tree-based model, which can effectively extract cross features, but also leverages convolutional neural networks to obtain more complex and inclusive features around the target location. Experimental results on real-world datasets demonstrate the superiority of our proposed LIMA model against the existing state-of-art methods. The LIMA model has been deployed as a web system for assisting one of the largest bank companies in Taiwan to select locations for building new branches in major cities since 2020.\u0000","PeriodicalId":45274,"journal":{"name":"ACM Transactions on Management Information Systems","volume":null,"pages":null},"PeriodicalIF":2.5,"publicationDate":"2024-04-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140731610","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Exploring How UK Public Authorities Use Redaction to Protect Personal Information","authors":"Yijun Chen, Reuben Kirkham","doi":"10.1145/3651989","DOIUrl":"https://doi.org/10.1145/3651989","url":null,"abstract":"\u0000 Document redaction has become increasingly important for individuals and organizations. This article investigates public-sector information redaction practices in order to determine if they adequately protect personal information from accidental disclosure due to redaction errors. Despite the importance of this in respect of data protection, 66.4% of those Public Authorities that responded did not hold formal policies or procedures\u0000 at all\u0000 . To assess those policies that did exist, we produced a 17-item check list of minimum best practice. Even those with policies and procedures had substantial defects to some degree (with the median performance being 29.4% on our checklist), with policies frequently recommending the use of high-risk redaction methods and overlooking essential practices. This means that these existing practices amount to widespread breaches of data protection law on the ground. To remedy this, we articulate a new set of document redaction standards, which overcome the existing inadequacies in current guidance, as well as make proposals for regulatory reform in this space.\u0000","PeriodicalId":45274,"journal":{"name":"ACM Transactions on Management Information Systems","volume":null,"pages":null},"PeriodicalIF":2.5,"publicationDate":"2024-03-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140251151","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Natalia Denisenko, Youzhi Zhang, Chiara Pulice, Shohini Bhattasali, Sushil Jajodia, Philip Resnik, V. S. Subrahmanian
{"title":"A Psycholinguistics-Inspired Method to Counter IP Theft using Fake Documents","authors":"Natalia Denisenko, Youzhi Zhang, Chiara Pulice, Shohini Bhattasali, Sushil Jajodia, Philip Resnik, V. S. Subrahmanian","doi":"10.1145/3651313","DOIUrl":"https://doi.org/10.1145/3651313","url":null,"abstract":"\u0000 Intellectual property (IP) theft is a growing problem. We build on prior work to deter IP theft by generating\u0000 n\u0000 fake versions of a technical document so that a thief has to expend time and effort in identifying the correct document. Our new\u0000 SbFAKE\u0000 framework proposes for the first time, a novel combination of language processing, optimization, and the psycholinguistic concept of surprisal to generate a set of such fakes. We start by combining psycholinguistic-based surprisal scores and optimization to generate two bilevel surprisal optimization problems (an Explicit one and a simpler Implicit one) whose solutions correspond directly to the desired set of fakes. As bilevel problems are usually hard to solve, we then show that these two bilevel surprisal optimization problems can each be reduced to equivalent surprisal-based linear programs. We performed detailed parameter tuning experiments and identified the best parameters for each of these algorithms. We then tested these two variants of\u0000 SbFAKE\u0000 (with their best parameter settings) against the best performing prior work in the field. Our experiments show that\u0000 SbFAKE\u0000 is able to more effectively generate convincing fakes than past work. In addition, we show that replacing words in an original document with words having similar surprisal scores generates greater levels of deception.\u0000","PeriodicalId":45274,"journal":{"name":"ACM Transactions on Management Information Systems","volume":null,"pages":null},"PeriodicalIF":2.5,"publicationDate":"2024-03-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140078485","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Giovanni Quattrocchi, Willem-jan Van Den Heuvel, D. Tamburri
{"title":"The Data Product-Service Composition Frontier: a Hybrid Learning Approach","authors":"Giovanni Quattrocchi, Willem-jan Van Den Heuvel, D. Tamburri","doi":"10.1145/3649319","DOIUrl":"https://doi.org/10.1145/3649319","url":null,"abstract":"\u0000 The service dominant logic is a base concept behind modern economies and software products, with service composition being a well-known practice for companies to gain a competitive edge over others by joining differentiated services together, typically assembled according to a number of features. At the other end of the spectrum, product compositions are a marketing device to sell products together in bundles that often augment the value for the customer, e.g., with suggested product interactions, sharing, etc. Unfortunately, currently each of these two streams—namely, product and service composition—are carried out and delivered individually in splendid isolation: anything is being offered as a product and as a service, disjointly. We argue that the next wave of services computing features more and more service fusion with physical counterparts as well as data around them. Therefore a need emerges to investigate the interactive engagement of both (data) products and services. This manuscript offers a real-life implementation in support of this argument, using (1) genetic algorithms (GA) to shape product-service clusters, (2) end-user feedback to make the GAs interactive with a data-driven fashion, and (3) a\u0000 hybridized\u0000 approach which factors into our solution an ensemble machine-learning method considering additional features. All this research was conducted in an industrial environment. With such a cross-fertilized, data-driven, and multi-disciplinary approach, practitioners from both fields may benefit from their mutual state of the art as well as learn new strategies for product, service, and data product-service placement for increased value to the customer as well as the service provider. Results show promise but also highlight plenty of avenues for further research.\u0000","PeriodicalId":45274,"journal":{"name":"ACM Transactions on Management Information Systems","volume":null,"pages":null},"PeriodicalIF":2.5,"publicationDate":"2024-02-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140423912","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Joaquin Delgado Fernandez, Tom Josua Barbereau, Orestis Papageorgiou
{"title":"Agent-based Model of Initial Token Allocations: Simulating Distributions post Fair Launch","authors":"Joaquin Delgado Fernandez, Tom Josua Barbereau, Orestis Papageorgiou","doi":"10.1145/3649318","DOIUrl":"https://doi.org/10.1145/3649318","url":null,"abstract":"With advancements in distributed ledger technologies and smart contracts, tokenized voting rights gained prominence within Decentralized Finance (DeFi). Voting rights tokens (aka. governance tokens) are fungible tokens that grant individual holders the right to vote upon the fate of a project. The motivation behind these tokens is to achieve decentral control within a decentralized autonomous organization (DAO). Because the initial allocations of these tokens is often un-democratic, the DeFi project and DAO of Yearn Finance experimented with a fair launch allocation where no tokens are pre-mined and all participants have an equal opportunity to receive them. Regardless, research on voting rights tokens highlights the formation of timocracies over time. The consideration is that the tokens’ tradability is the cause of concentration. To examine this proposition, this paper uses an agent-based model to simulate and analyze the concentration of voting rights tokens post three fair launch allocation scenarios under different trading modalities. The results show that regardless of the allocation, concentration persistently occurs. It confirms the consideration that the ‘disease’ is endogenous: the cause of concentration is the tokens’ tradablility. The findings inform theoretical understandings and practical implications for on-chain governance mediated by tokens.","PeriodicalId":45274,"journal":{"name":"ACM Transactions on Management Information Systems","volume":null,"pages":null},"PeriodicalIF":2.5,"publicationDate":"2024-02-23","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"140435531","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Design with Simon's Inner and Outer Environments: Theoretical Foundations for Design Science Research Methods for Digital Science","authors":"V. Storey, Richard Baskerville","doi":"10.1145/3640819","DOIUrl":"https://doi.org/10.1145/3640819","url":null,"abstract":"Design science research has traditionally been applied to complex real-world problems to produce an artifact to address such problems. Although design science research efforts have been applied traditionally to business or related problems, there is a large set of problems in the area of digital science that also require important, digital artifacts. The digitalization of science has resulted in the need to develop essential, specialized, devices and software before it is feasible for scientists to carry out their work. This research examines digital science to identify its challenges and demonstrate how it can be possible to progress digital science with design science research, thereby establishing digital science as an important area of transdisciplinary inquiry. These areas of research are examined for their synergies and explained by positioning artifact development challenges with respect to Simon's inner and outer environments, and the interface between them.","PeriodicalId":45274,"journal":{"name":"ACM Transactions on Management Information Systems","volume":null,"pages":null},"PeriodicalIF":2.5,"publicationDate":"2024-01-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"139619044","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}