Blood Bank Inventory Management to Handle Demand and Disruptions due to COVID19
Collaborators: Calvino Cheng, Jason Quinn, Robert Liwski (Dept. of Pathology and Laboratory Medicine, NSHA)
Synopsis: Due to the COVID-19 pandemic, there is an increased concern around shortages of blood products due to impacts on the availability of blood donors; this could seriously affect the operations of hospitals and, in turn, patient safety across Canada. In response to potential blood product shortages, this project is developing a data-driven blood product inventory management solution to: (1) visualize on an interactive dashboard the near real-time inventory and on-going transactions of blood products, across the health system, to anticipate the demand and, in turn, streamline the distribution of essential blood products; (2) prevent the expiration of blood products by appropriate recycling between hospitals by transferring a near outdated blood product from one hospital to another hospital that is in need of the product by avoiding imminent wastage. The project’s impact will be the effective management of blood products so that essential healthcare services can be delivered despite supply disruptions due to COVID-19.
Improving Deceased Donor Kidney Allocation Using Machine Learning Methods to Develop a Survival Prediction Model
Collaborators: Karthik Tennankore, Amanda Vinson (Dept. of Nephrology, NSHA)
Synopsis: Kidney transplantation is the optimal form of renal replacement therapy, but patients are still at risk of death or premature kidney graft loss. Studies have identified that characteristics of the donor kidney (including immunological and non-immunological factors) and recipient characteristics (including demographics, comorbid conditions and immunological factors) at the time of transplant influence short and long term outcomes after transplantation. Risk stratification models to determine optimal organ allocation, such as Kidney Donor Risk Index and the Estimated Post Transplant Survival score, are not too effective at discriminating between outcomes across the entire population of recipients. In this project, we are investigating machine learning methods to develop and validate a kidney donor and recipient risk prediction tool for death, graft loss and the combination of death and graft loss after kidney transplantation, using immunological and non-immunological paired donor recipient variables, to achieve optimal organ allocation, and provide insights into measures of the important post-transplant outcomes of death, and graft loss.
A Digital Health Eco-System to Deliver Personalized Rehabilitation Therapy to Treat Pediatric Neurodevelopmental Disorders in Brazil
Collaborators: Raquel Dias, Marcela de Oliveira Lima, João Guilherme Bezerra Alves, Rehabilitation Centre Prof. Fernando Figueira Integral Medicine Institute, Recife, Pernambuco, Brazil
Synopsis: The recent outbreak of the Zika virus in Brazil has caused microcephaly in newborns, which affects cognitive development delays. Treatment for developmental delays in children between 0-3 years comprises a standardized exercise program—known as the Early Stimulation Program (ESP) that targets cognitive and motor function development. In Brazil, access to ESP therapy is available at a limited number of specialized therapeutic centers, meaning that many affected children do not get appropriate treatment in a timely manner. To improve timely access to ESP treatment, we are implementing a home-based ESP delivery approach that remotely trains the parents of affected children, by delivering to them short videos on their mobile phone, about how to perform the ESP at home. In this project, we are developing, implementing and evaluating a digital health based ESP therapy intervention called the BraziLian Early Stimulation System (BLESS) that incorporates (a mobile health app that allows parents to receive the prescribed ESP videos and associated educational material help them administer ESP in a home-based setting, and (b) an intelligent ESP therapy planning and decision support tool to support therapists to design an evidence-based ESP based on the child’s needs. BLESS is being deployed at a reference center for Maternal and Child healthcare in the Northeast of Brazil, where there is a high incidence of developmental delays in children. An evaluation study will be conducted to measure the efficacy of our approach.
Acromegaly Facial Features – Novel Strategies Comparing Patients, Specialists And Computerized Facial Recognition Software
Collaborators: Syed Ali Imran (Dept. of Endocrinology, NSHA)
Synopsis: Acromegaly is a slowly developing disease due to high growth hormone levels released from a benign pituitary tumor that causes many complications such as severe facial disfigurement, enlargement of hands and feet, heart disease, cancer, sleep apnea, severe psychological problems and early death. In this project we aim to investigate a novel acromegaly screening strategy that asks non-specialists (particularly by patients themselves) to detect the presence of acromegaly using an acromegaly screening tool. We are developing a patient-centered home-based acromegaly screening mobile app that will enable patients to self-grade their photograph using the screening tool to detect the presence of acromegaly. We are investigating the use of facial recognition technology to grade the patient’s photograph to achieve automated acromegaly screening.
Analyzing Blood Product Transactional Data Using Machine Learning Methods for Reducing Blood Unit Wastage
Collaborators: Calvino Cheng, Jason Quinn (Dept. of Pathology and Laboratory Medicine, NSHA)
Synopsis: Blood products are expensive and perishable thus requiring efficient and effective inventory management practices to minimize wastage as it seriously impacts the efficiency and appropriate use of blood products. This project aims to reduce blood product wastage due to discards by (a) identifying the underlying patterns leading to blood product discards by analyzing institutional blood transactional data using state-of-the-art AI based machine learning methods; (b) applying the insights about the blood product’s lifecyle to proactively detect a potential blood product that is suseceptible of being discarded based on its lifecycle. Using machine learning methods, we are analyzing a large high-resolution transactional dataset of 5 years (2014-2019; approximately 1,200,000 records) for all blood products from the Laboratory Information System at the Halifax Central Zone. The project will deliver an innovative data-driven solution to analyze blood product transactions to proactively detect potential blood units that are likely to be wasted.
Advance Data Analytics and Visualization using Pathology Test Ordering Data to Optimize Pathology Laboratory Utilization
Collaborators: Calvino Cheng, Manal Elnenaei, Bryan Crocker (Dept. of Pathology and Laboratory Medicine, NSHA), Samuel Campbell (Dept. of Emergency Medicine, NSHA)
Synopsis: The ‘Choosing Wisely Canada’ movement is promoting the provision of sustainable healthcare by optimizing the utilization of healthcare services and adhering to best practices in clinical decision making. It has been observed that a significant number of pathology tests ordered by physicians are inappropriate, leading to suboptimal laboratory utilization and patient safety. In this project, we are employing big’ data analytics, machine learning and data visualization methods to develop a laboratory utilization management platform termed as Pathology Laboratory Utilization Scorecards (PLUS) to provide test ordering insights in terms of (i) interactive scorecards for physicians to examine their test ordering pattern across different tests and over time, rate of inappropriate tests, compliance with clinical guidelines and how they compare with peers, and (ii) interactive dashboards for laboratory managers to assist them with resource planning and waste minimization. PLUS will be implemented to optimize the central zone pathology laboratory in Halifax (the largest provincial lab) that on average processes 8 million general lab orders.
Using Predictive Analytics for Stratification of Surgical Patients in Orthopaedics to Reduce Surgery Wait times and Improve Surgical Outcomes.
Collaborators: Michael Dunbar, Kathryn Young-Shand (Dept. of Orthopaedics, NSHA), Janie Wilson (Dept. of Biomedical Engineering, Dalhousie University)
Synopsis: Reducing the wait lists for Total Knee Arthroplasty (TKA) whilst improving surgical outcomes is a major consideration for the Nova Scotia healthcare system, and even across Canada. One approach to improve surgical outcomes is to objectively select TKA recipients with the greatest need and improvement potential. To better screen candidate patients, we are investigating how a combination of preoperative and postoperative clinical and biomechanical features correlate with objective surgical outcomes. The predictive analytics aspect of this research involves the use of machine learning methods to (a) identify relevant patient phenotypes and clinical biomechanical phenotypes among a heterogeneous group of TKA candidates with differing demographics and gait biomechanics, (b) quantify and visualize disease severity and individual biomechanical variability among asymptomatic and TKA candidates, (c) abstract decision rules to classify patients among the relevant patient severity phenotypes prior to arthroplasty, and (d) assess how patient phenotype prior to surgery relates to outcomes after arthroplasty. This work is being conducted in collaboration with the School of Biomedical Engineering at Dalhousie.
Using AI-based Speech Recognition for Digital Point-of-care Clinical Documentation in a Busy Pediatric Emergency Department
Collaborator: Brett Taylor (Dept. of Pediatric Emergency, IWK)
Synopsis: Completing clinical documentation (i.e. patient charts) during an encounter is quite time-consuming and many physicians capture cursory clinical notes leading to sub-optimal recording of the encounter. In an Emergency Department (ED), preparing complete patient charts is even more difficult given the intense nature of ED with physicians attending multiple patients in short time intervals. In this project, we are developing an Intelligent Clinical Dictation and Transcription System (ICDTS), that will allow physicians to directly narrate the clinical encounter without interrupting their care workflows, and, in turn, the system will produce a structured electronic patient chart. The resulting structured patient chart can be analyzed using data mining methods for clinical research and process management (patient pathways, physician output, admission and return rates). We are applying machine learning methods to a corpus of clinical charts to train acoustic and language models tailored to the ED setting to construct a speech-to-text engine with near human accuracy on continuous medical speech. We will evaluate the performance of ICDTS by conducting a pilot study at the IWK PED with ED physicians using ICDTS in real-life Pediatric ED settings.
Ambient Assisted Living Framework to Assist Patient Self-Management Activities
Assisted Ambient Living (AAL) focuses on improving self-sufficiency of individuals, especially individuals living with disabilities or chronic conditions, in terms of their ability to perform activities of daily life (ADL) and self-managing their condition in a home-based setting. Activity recognition for AAL, involves inferring an individual’s ongoing activities based on personal and environment sensors. Activity recognition is a pivotal task in AAL, as it enables the AAL system to understand the context and state of the individual and then accordingly guide the individual to perform ADL or issue alerts in case of unusual activities are detected.
The objective of this project is to investigate semantic web based methods to develop a knowledge-based activity recognition framework that can detect specific activities of an individual in an AAL environment. The intent of activity recognition is to support individuals to (a) perform ADL activities to self-manage their condition through context-sensitive guidance, and (b) change unhealthy behaviours through timely personalized messages in response to observed unhealthy activities. We are investigating semantic web based methods to (i) develop high-level ontological activity models that entail contextual and temporal aspects of activities, coupled with activity related sensors, to represent activities; (ii) develop ontological models to represent an individual’s behaviour towards completion of specific activities; (iii) develop behaviour modification strategies, building on inferred activities, to assist individuals undergo behaviour modification to efficaciously self-manage their condition.
We have developed a prototype AAL system to monitor activities related to medication adherence. Medication adherence is monitored using a smart pillbox that sends events to a smartphone, and smart home sensors to detect patient movement and the use of objects (e.g. cup or glass) for taking medicines. To improve medication adherence, patients and/or their care providers are alerted in case a scheduled medication activity is not detected. Our AAL approach is non-obtrusive and it maintains the privacy of the household members as it only detects activities related to taking medications. We are investigations into extending the activity recognition model and its application to other home-based patient care activities.
Research Areas (Computer Science): Activity recognition, Semantic web, Knowledge modeling, Machine learning, Sensor data fusion
Research Areas (Health): Assisted living, Behaviour modification, Self-management, Home-based care
Project Duration: 2016 – ongoing
PLUS: Pathology Laboratory Utilization Scorecards
The ‘Choosing Wisely Canada’ movement is promoting sustainable healthcare by optimizing the utilization of healthcare services. Pathology laboratories provide service to help with disease diagnosis and therapeutic choices. A meta-analysis of 108 studies involving 1.6 million results from 46 of the 50 most commonly ordered lab tests, found that on average 30 % of all tests are likely to be unnecessary. The question pursued in this project is whether pathology test ordered in Nova Scotia are relevant and useful for patient management?
This project is in collaboration with Department of Pathology, and its objective is to optimize laboratory utilization from a primary care perspective by: (a) informing primary care physicians about their test ordering profile to help them reduce laboratory overutilization; (b) providing laboratory managers with intelligence about laboratory utilization patterns, to help them minimize waste.
In this health data analytics project, we are investigating and developing a digital health based Pathology Laboratory Utilization Scorecard (PLUS) system that leverages advanced data analytics and data visualization techniques to analyze ‘big’ laboratory data to generate objective, descriptive and predictive laboratory utilization insights to optimize laboratory utilization. We are applying advance data analytics methods to ‘learn’ physician-specific laboratory utilization models derived from their laboratory utilization data. To display laboratory utilization, we use visual analytics techniques to display an interactive and dynamic (i) physician scorecard for physicians to self-audit their test ordering patterns; and (ii) laboratory manager dashboard to illustrate laboratory utilization for resource planning. PLUS is a web-based system that after evaluation will be used to optimize the pathology laboratory in the central zone (Halifax) that annually performs on average 15 million laboratory tests for 200,000 patients.
Research Areas (Computer Science): Big data analytics, Machine learning, Data visualization, Dashboards, Data fusion
Research Areas (Health): Pathology, Choosing wisely, Health system use optimization
Project Duration: 2017 – 2019
PRISM: Personalized Risk Investigation, Stratification and Mitigation Platform
Chronic non-communicable diseases, such as cardiovascular disease, diabetes and cancers are regarded as lifestyle-related diseases. Lifetime health is an emerging health paradigm that aims to assist individuals to achieve long-term health targets, and avoid harmful lifecycle choices to mitigate the risk of chronic diseases. Public Health Agency of Canada’s strategic directions indicate the urgent need for innovative citizen-engagement and empowerment strategies to preemptively detect and effectively prevent the onset of chronic diseases.
The PRISM platform is an innovative population health oriented lifetime health eco-system to empower Canadian to take personalized measures to prevent the onset of chronic diseases by performing periodic risk assessment, monitoring, lifestyle change to pursue lifetime health.
The PRISM platform, currently being developed, provides informative and interactive tools for individuals to (a) assess their cumulative chronic disease risk by providing their health status, and (b) pursue personalized behavioral interventions targeting lifestyle changes to mitigate the noted risk of chronic diseases. At present, PRISM assess risks for 4 chronic conditions and 8 cancers. Additionally, PRISM integrates the individual chronic disease risk scores to provide a cumulative Health Asset Score. The PRISM platform leverages artificial intelligence, data visualization and mobile health technologies, and it is accessible via web and as a mobile app.
Research Areas (Computer Science): Semantic web, Knowledge modeling, Data analytics, Data visualization
Research Areas (Health): Lifetime health, Chronic disease risk assessment, Disease prevention, Behaviour modification, Population/public health, Education and empowerment, Personalized health
Project Duration: 2017 – ongoing
Engage: Behaviour Modification for Chronic Illness
Lifetime healthcare is an emerging health paradigm, aiming to assist individuals achieve health targets and avoid harmful lifestyle choices to lead a healthy life. Behavior modification is an integral element of lifetime healthcare—personalized behavior modification interventions help individuals to abandon health-compromising behaviors and adopt health-enhancing behaviors.
This behaviour modification project aims to assist chronic disease patients (such as diabetes) to achieve self-management efficacy by helping them overcome barriers to healthy lifestyles. We are developing a personalized behavior modification framework, called Engage, that (a) incorporates the Social Cognitive Theory (SCT) to generate personalized behaviour modification action plans, (b) uses consumer health devices to collect health data in a home-setting to monitor outcomes related to positive health behaviours, and (c) employs social modeling techniques to encourage individuals to pursue their behaviour modification plans.
We are investigating a knowledge-based, action-planning and community-driven approach to help patients achieve two key SCT targets—i.e. knowledge and self-efficacy to achieve positive behaviour modification. Our approach is to formulate personalized behavior modification programs as sequences of short-term action plans, targeting to overcome the individual’s perceived barriers to long-term behavior change. We are investigating semantic web methods to formally represent SCT knowledge and to digitize the behaviour modification content in terms of a high-level ontological SCT knowledge model. Reasoning over the SCT knowledge model allows to generate personalized action plans as per the patient’s health and behavioral profile. We have investigating a social modelling approach to (a) help patients select action plans by showing the success rate and experiences of similar patients within the Engage community; (b) offer motivation to patients to share the success rate of similar patients; and (c) encourage patients to connect, exchange advice and provide encouragement on action plans.
The Engage framework has been implemented and is accessible to participating patients through the Engage mobile health app and a web-based system. Engage offers an innovative approach to behavior modification that includes is personalized to an individual user but also builds on the collective experience and outcomes of a larger community of patients undergoing behaviour modification interventions.
Research Areas (Computer Science): Semantic web, Knowledge modeling, Social computing, Mobile computing
Research Areas (Health): Personalized health, Behaviour modification, Home-based care
Project Duration: 2017 – ongoing