Clinical Pathways for Diagnosing Neurocognitive Disorders: Insights From Process Mining a Memory Clinic Cohort.

BackgroundSeveral clinical guidelines and recommendations have been developed to improve the diagnostic process of patients with dementia, with the aim of optimising patient care and fostering a cost‐effective use of resources. However, patient characteristics and organizational factors in clinical practice can lead physicians to deviate from recommendations. In this context, process mining (PM) is a powerful technique for evaluating the compliance of actual to ideal diagnostic pathways, but has never been used in the cognitive field. This study aims to pilot the use of PM to study the diagnostic workup of patients with cognitive complains in an academic memory clinic.MethodWe retrospectively reviewed medical charts of 539 consecutive new patients referred to the University Hospitals of Geneva (Switzerland) between July 2021 and December 2022. The collected information includes socio‐demographic data, number and dates of clinical visits and biomarker exams (e.g., FDG‐, amyloid‐, tau‐PET, cerebrospinal fluid ‐ CSF). The pMineR package, a R library for PM, was used for data inspection and process discovery (e.g., First Order Markov Model ‐ FOMM, Careflow Miner ‐ CFM) to extract features of most frequent diagnostic pathway according to clinical stage.ResultPatients were 96 subjective cognitive decline (SCD; MMSE: 28.5±1.6), 308 Mild Cognitive Impairment (MCI; MMSE: 26.5±2.7), and 135 Mild Dementia (DEM; MMSE: 22.7±3.2). FOMM showed that all SCD patients received a syndromic diagnosis based on a detailed neuropsychological evaluation and a morphological imaging. CSF was the first‐line biomarker to ascertain the aetiology of cognitive impairment, used in 90 MCI (30%) and 32 DEM (24%). A second‐line biomarker (e.g., amyloid‐, FDG‐, Tau‐PET) was required in 23% of MCI and 16% of DEM. Differential PM with CFM revealed that DEM was more commonly referred to the first clinical assessment with morphological imaging than MCI (DEM:68/135vs.MCI:123/308, ratio:1.8; p = 0.047).ConclusionThese preliminary results demonstrate that the PM approach is appropriate for describing the diagnostic workup in memory clinics. Using the same approach and study design, a further project will evaluate the diagnostic pathways in 6 European academic memory clinics and contrast them to recommendations recently issued from a large European consensus effort (Frisoni, LancetNeurol 2024).

Objectives: Diagnostic pathways are limited. A validated instrument that can triage patients when they are suspected of mild dementia (MD) is necessary to optimise referrals. Method: The MoCA is validated for identifying MD and mild cognitive impairment (MCI) in a cohort of patients suspected of cognitive impairment (CI) after initial assessment in old age psychiatry. The reference standard was the consensus-based diagnoses for MD and MCI, adhering to the international criteria and using suspected patients, but without CI as comparisons (NoCI). Results: The mean MoCA scores differ significantly between the groups: 24(SE: .59) in NoCI, 21(SE: .31) in MCI and 16,7(SE: .45) in MD (p < .05). The AUC of MD against non-demented (MCI + NoCI) was 0.83(95%CI: 0.78–0.88) resulting in 90% sensitivity, 65% specificity, 50%PPV and 94%NPV at a “best” cutoff of <21 according the Youden index and respectively 0.77(95%CI: 0.69–0.85), 56%, 73%, 90%, 28% for CI (MD + MCI) against NoCI at <21. Conclusion: 90% of individuals with a MoCA of <21 will have CI (MD + MCI), while 94% with a MoCA of ≥21 will not have dementia. The MoCA can reduce referrals substantially (50%) by selecting who don’t need further work up in a memory clinic, even if they were suspected of CI after initial assessment.

Process mining in healthcare: A literature review

In the context of online banking, new users have to register their information to become clients through mobile applications; this process is called digital onboarding. Fraudsters often commit identity fraud by impersonating other people to obtain access to banking services by using personal data obtained illegally and causing damage to the organisation’s reputation and resources. Detecting fraudulent users by their onboarding process is not a trivial task, as it is difficult to identify possible vulnerabilities in the process to be exploited. Furthermore, the modus operandi for differentiating the behaviour of fraudulent actors and legitimate users is unclear. In this work, we propose the usage of a process mining (PM) approach to detect identity fraud in digital onboarding using a real fintech event log. The proposed PM approach is capable of modelling the behaviour of users as they go through a digital onboarding process, while also providing insight into the process itself. The results of PM techniques and the machine learning classifiers showed a promising 80% accuracy rate in classifying users as fraudulent or legitimate. Furthermore, the application of process discovery in the event log dataset produced an insightful visual model of the onboarding process.

Process mining has emerged as a way to analyze systems and their actual use based on the event logs they produce. The focus of process mining, unlike other business intelligence domains, is on concurrent processes and not on static or mainly sequential structures. Process mining is applicable to a wide range of systems. These systems may be pure information systems (e.g., ERP systems) or systems where the hardware plays a more prominent role (e.g., embedded systems). The only requirement is that the system produces event logs, thus recording (parts of) the actual behavior. Current Process Mining Algorithms (PMAs) face two major challenges: 1) real-life event logs contain large amounts of data about previous executions of a process, and 2) since they attempt to derive accurate information from the event logs, PMAs are computational expensive. Moreover, process mining experts often combine multiple PMAs to offer insights into systems real behavior from different perspectives, i.e. they execute process mining workflows. These workflows are currently executed manually sequentially or are hard-coded. In the past decade, new emerging concepts such as grid computing, service-oriented architectures and cloud computing provide solutions to the increasing demand for data storage and computing power. These technologies enable worldwide distributed resources, e.g. software and infrastructure, to cooperate for a specific user defined goal. Such distributed environments can, on one hand, offer a solution to the complexity challenges of the process mining domain and, on the other hand, create the possibility to enable PMAs as services that can be combined and orchestrated via workflow engines. This PhD thesis proposes a framework for the execution of process mining workflows in a distributed environment. The distribution of the PMAs is done at two levels: 1) process mining algorithms are parallelized, thus reducing considerably their time consumption; and 2) a framework for automated execution of process mining workflows is proposed. For the first level, we focus on one particular advanced process mining algorithm - Genetic Mining Algorithm (GMA), and we propose two distribution strategies Distributed Genetic Mining Algorithm (DGMA) and Distributed Sample-based Genetic Mining Algorithm (DSGMA), improving significantly the GMA time efficiency. DGMA distributes the GMA computation on different computational resources by using a coarsegrained approach. The second strategy, DSGMA further reduces the computation time by data distribution and exploiting the event logs redundancy. For both of the algorithms, we derive guidelines for their parameter configuration based on empirical evaluations; we validate the guidelines on several real life event logs. All the proposed algorithms described in this thesis have been implemented as plug-ins in the ProM framework - an open source tool available at www.processmining.org. For the second level, we provide a formal description of a grid architecture suitable for process mining experiments in terms of a colored Petri net (CPN). The CPN can be seen as a reference model for grids and clarifies the basic concepts at a conceptual level. Moreover, the CPN allows for various kinds of analysis ranging from verification to performance analysis. The level of detail present in the CPN model allows us to, almost straightforwardly; implement a real grid architecture based on the model. Note that even if our reference model was inspired by the challenges from process mining domain challenges, it can be used for other computationally challenging domains as well. Based on the CPN reference model we implemented a prototype, called YAGA (Yet Another Grid Architecture), a service-based framework for supporting process mining experiments. YAGA is a simple, but extensible, grid architecture that combines a powerful workflow engine YAWL and the ProM framework through a JAVA-based grid middleware. By combining the CPN reference model and YAGA, we provide a powerful grid framework for process mining experiments. The model allows for easy experimentation and extensive debugging. It also ensures an easy and rapid way to choose optimal parameters of real life workflows. These estimations can help users in planning their experiments and/or re-configuring their workflows. Moreover, performing model simulations on-the-fly can give realistic resource load predictions, which can be used for improving the scheduling process.

Semantics has been a major challenge when applying the process mining (PM) technique to real-time business processes. The several theoretical and practical efforts to bridge the semantic gap has spanned the advanced notion of the semantic-based process mining (SPM). Fundamentally, the SPM devotes its methods to the idea of making use of existing (semantic) technologies to support the analysis of PM techniques. In principle, the semantic-based process mining method is applied through the acquisition and representation of abstract knowledge about the domain processes in question. To this effect, this paper demonstrates how the semantic concepts and process modelling (reasoning) methods are used to improve the outcomes of PM techniques from the syntactic to a more conceptual level. To do this, the study proposes an SPM-based framework that shows to be intelligent with a high level of semantic reasoning aptitudes. Technically, this paper introduces a process mining approach that uses information (semantics) about different activities that can be found in any given process to make inferences and generate rules or patterns through the method for annotation, semantic reasoning, and conceptual assertions. In turn, the method is theoretically applied to enrich the informative values of the resultant models. Also, the study conducts and systematically reviews the current tools and methods that are used to support the outcomes of the process mining as well as evaluates the results of the different methods to determine the levels of impact and its implications for process mining.

Process mining is an analytical approach which stems from and converges on data science and process modelling. Initially incepted to support business process management, however process mining approach is universal and applicable to other fields. It was already discerned that process mining techniques share similarities with such used in bioinformatics and that the emerging process mining discipline can benefit from applying techniques developed in computational biology [1]. Herein however, we demonstrate the reverse: that process mining can be applied for the study biological processes. As process mining operates on event logs in order to analyze a particular biological process it is necessary to transform the information for a sequence of biological events into an event log. For this study we applied process mining techniques to a developmental dataset from the lineage-resolved molecular atlas of the round worm C. elegans [2]. The single-cell temporal gene expression data was transformed into event log and analyzed with process mining tools. We show that application of process mining to biological processes is feasible, yet the presentation of the results with current tools is not suitable for the high information content of the particular biological process and this hampers further extraction of knowledge. We conclude that the application of process mining to biological processes would be beneficial for both fields.

Pieces of literature discussing the process model in the learning management system are limited to student and teacher learning interactions. Including the learning interactions of principals and parents contributes to more detail of processes taking place during learning interactions on the platform. The study used process mining techniques and algorithms to extract the underlying processes that drive learning interactions in social learning management systems. The discovered processes for principals, teachers, students, and parents consequently show a precision value of 1, 0.542, 0.639, and 1, respectively. The preciseness of processes for each user group indicates an acceptable behavior (> 0.50) extracted from the event logs. On the other hand, social networks form from the processes that show the information flow of learning interactions from the principal to the students and parents, depicting everyone's effort for learning gain in favor of the student. This study's contribution expands beyond teacher-student interaction processes to include principals and parents, thereby generating a more concrete view of learning interaction in the social learning management system.

Among the knowledge areas in which process mining has had an impact, the audit domain is particularly striking. Traditionally, audits seek evidence in a data sample that allows making inferences about a population. Mistakes are usually committed when generalizing the results and anomalies; therefore, they appear in unprocessed sets; however, there are some efforts to address these limitations using process-mining-based approaches for fraud detection. To the best of our knowledge, no fraud audit method exists that combines process mining techniques and visual analytics to identify relevant patterns. This paper presents a fraud audit approach based on the combination of process mining techniques and visual analytics. The main advantages are: (i) a method is included that guides the use of the visual capabilities of process mining to detect fraud data patterns during an audit; (ii) the approach can be generalized to any business domain; (iii) well-known process mining techniques are used (dotted chart, trace alignment, fuzzy miner…). The techniques were selected by a group of experts and were extended to enable filtering for contextual analysis, to handle levels of process abstraction, and to facilitate implementation in the area of fraud audits. Based on the proposed approach, we developed a software solution that is currently being used in the financial sector as well as in the telecommunications and hospitality sectors. Finally, for demonstration purposes, we present a real hotel management use case in which we detected suspected fraud behaviors, thus validating the effectiveness of the approach.

Semantics has been a major challenge when applying the Process Mining (PM) technique to real-time business processes. In theory, efforts to bridge the semantic gap has spanned the advanced notion of Semantic-based Process Mining (SPM). The SPM devotes its methods to the idea of making use of existing semantic technologies to support the analysis of PM techniques. Technically, the semantic-based process mining is applied through acquisition and representation of abstract knowledge about the domain processes in question. To this effect, this paper demonstrates how semantically focused process modelling and reasoning methods are used to improve the outcomes of PM techniques from the syntactic to a more conceptual level. Also, the work systematically reviews the current tools and methods that are used to support the outcomes of the process mining, and to this end, propose an SPM-based framework that proves to be more intelligent with a higher level of semantic reasoning aptitudes. In other words, this work provides a process mining approach that uses information (semantics) about different activities that can be found in any given process to generate rules and patterns through the method for annotation, conceptual assertions, and reasoning. Moreover, this is done to determine how the various activities that make up the said processes depend on each other or are performed in reality. In turn, the method is applied to enrich the informative values of the resultant models.

Visual analytics (VA) integrates the outstanding capabilities of humans in terms of visual information exploration with the enormous processing power of computers to form a powerful knowledge discovery environment. In other words, VA is the science of analytical reasoning facilitated by interactive interfaces, capturing the information discovery process while keeping humans in the loop. Process mining (PM) is a data-driven and process centric approach that aims to extract information and knowledge from event logs to discover, monitor, and improve processes in various application domains. The combination of interactive visual data analysis and exploration with PM algorithms can make complex information structures more comprehensible and facilitate new insights. Yet, this combination remains largely unexplored. In this article, we illustrate the concepts of VA and PM, how their combination can support the extraction of more insights from complex event data, and elaborate on the challenges and opportunities for analyzing process data with VA methods and enhancing VA methods using PM techniques.

P185: Comparison of social function in mild cognitive impairment and mild dementia using the Japanese version of the Social Functioning in Dementia scale (SF-DEM-J)

Tests measuring proactive semantic interference as The Loewenstein-Acevedo Scales for Semantic Interference and Learning (LASSI-L), has shown promising diagnostic properties for the diagnosis of Mild Cognitive Impairment (MCI) and dementia. LASSI-L may also be efficient in predicting cognitive decline in at-risk individuals. There is an unmet need to examine the diagnostic properties of the LASSI-L in a Danish context where traditional neuropsychological tests are typically applied when diagnosing possible dementia disorders. To investigate the reliability, convergent validity, and predictive value of the new Danish LASSI-L version in aMCI and mild dementia due to Alzheimer’s disease (AD). From a memory clinic we included 17 aMCI patients, 15 patients with mild dementia (AD), 17 patients with Subjective Cognitive Decline (SCD), and 30 healthy controls. Neuropsychological assessment was applied in all patients, and biomarker analyses were performed for patients with aMCI and mild AD. Cronbach’s alpha was 0.94. Patients with aMCI and mild dementia differed significantly from healthy controls on all LASSI-L measures. ROC analyses showed a very high AUC value for both patients with aMCI [0.85–0.97] and mild dementia [0.93–0.99]. SCD patients generally did not differ from controls, except for significantly lower scores on one item (Cued Recall A1) LASSI-L had high reliability and promising predictive value in the diagnosis of aMCI and mild AD due to AD. SCD patients diagnosed in a memory clinic did not differ significantly from healthy on the LASSI-L.

Business process mining or process mining is the intersection between data mining and business process modelling that extracts business patterns from event logs. Event logs are freely available in any organization. Business logs are a potential source of useful information. By the various patterns that are present in the logs, a lot can be estimated about the type of procedures that should be incorporated into the organization for better performance. Event logs store information about time and event data of business processes. Process mining algorithms are used to mine business process models using event logs. Generating automated business models out of this could provide valuable insight to a firm eventually leading to customer satisfaction. Process Mining works by three phases: discovery, conformation and alteration. By using process mining, many kinds of information can be collected about the process, such as control-flow, performance, organizational information and decision patterns. A process model could be represented as Petri nets which is a formal graphical representation of the workflow diagram or it can be represented as Business Process Modelling Notation. This project aims to develop a user friendly platform which is capable of generating petri net like models by process mining. By using various process mining algorithms we will develop software which would mine the event logs of a particular firm. It would provide a data or workflow analysis scheme. This would optimize business process intelligence and thus provide alternative and superior work strategies. In this project, we are mainly targeting project management using process mining. There are many projects that are undertaken by an IT company that all follow the same procedure. The concept of business process mining can be used in order to improve the performance of a company by optimizing its Software Development Life Cycle. By feeding the previous logs of a similar project of the company, the software would give a flowgraph. This flowgraph can help to identify the sequence of the activities, roles in the organization as well as various efficiency parameters. The algorithm being used is the Heuristic Miner Algorithm for process mining.

Background: Despite the increasing burden of dementia in developing countries, mild cognitive impairment (MCI) continues to be underexplored. MCI has conventionally been identified based on clinical profile, but recently, biomarkers suggestive of Alzheimer's disease pathology have been included in the revised National Institute on Aging and the Alzheimer's Association (NIA-AA) criteria. In this study, we evaluated the profile of MCI in a memory clinic in India and explored the applicability of the revised NIA-AA criteria in a limited resource setting. Methods: Consecutive subjects evaluated at the memory clinic for mild memory complaints were included and underwent clinical and neuropsychological examination as well as standard brain imaging. A subset of patients was subjected to imaging biomarker studies as a part of routine clinical practice. Results: Among the 1,190 patients evaluated during the study period, 226 (19.0%) presented with mild memory complaints. Cerebrovascular disease was a common secondary cause. Nearly half of the patients (109 of 226) had MCI according to the modified Petersen criteria. All MCI subjects were educated and the majority were male. A total of 12% of the cohort was classified by imaging biomarkers as having MCI with intermediate likelihood of AD according to the NIA-AA criteria. Conclusion: In the setting of urban India, MCI is an emerging problem; therefore, it was feasible to operationalise the revised NIA-AA criteria in identifying subjects with MCI with intermediate likelihood of AD.