Brisbane 2018 problems
Matching people with great psychologists: Automatically recommending suitable psychologists for people based on their needs, preferences, and demographics.
Many people have trouble finding a psychologist they ‘click’ with. Millions of Australians never seek the help they need, and many of those who do give up after one session, or waste money and emotional energy ‘auditioning’ multiple psychologists.
Our service currently makes automated recommendations for some website visitors with human clinicians contributing guidance and recommendations about half the time. The recommender system is hand-tuned. The amount of human involvement makes it hard to serve the large number of people looking for help.
Epilepsy sufferers regard the unpredictability of seizures as the main distress. Being able to predict when a seizure will occur would largely reduce the fear these patients feel when carrying out normal activities such as being in a public space or driving a car. There is no wearable device that can produce reliable real-time alarms.
The OpenHeart Project (https://openheartproject.org) , currently lead by the Innovative Cardiovascular Engineering and Technology Laboratory (ICETLAB), is an open-source online research project which aims to improve research practices within the field of mechanical circulatory support and ultimately improve outcomes and quality of life of heart failure patients around the globe. OpenHeart Project currently maintains a collaborative platform to share designs and testing for mechanical circulatory support.
To complement this, the OpenHeart Project would like to curate and display dataset around heart failure and mechanical circulatory support.
Collating this data will open up multiple opportunities for new research and improved outcomes. As an immediate short-term translation the database can function as an educational tool for students, researchers and the general public interested or working in the field of mechanical circulatory support. Furthermore, the database can be a starting point to consider big data / machine learning and model predictive tools to predict future trends in mechanical circulatory support.
The advent of real-time video chat has enabled people to communicate naturally regardless of their location. One group of people, however, have not experienced quite the same level of empowerment. People who can only communicate using sign language are limited to text chat when they need to communicate remotely, a medium with clear limitations. Perhaps most importantly, with the growing adoption of telehealth, deaf people need to be able to communicate naturally with health practitioners, regardless of whether the practitioner knows sign language.
50% of Australians suffer from one or more chronic diseases and up to 70% of people forget what information their doctors told them post consultation within 15 minutes of leaving the clinic. As the Australian population ages, there is an increasing need to support the existing and struggling health infrastructure by providing valuable digital health interventions. No platform in Australia is publicly available which supports the ageing population to manage their medication through an SMS based service, which research suggests can provide beneficial adherence outcomes. We also understand that medication management is not only a problem of forgetfulness, rather a multivariate problem which can be encompassed by behaviours, beliefs and barriers.
Despite all the advances in technology, healthcare continues to rely on paper, fax machine, internal mail and pager. These current modes of communication are simply inefficient, insecure, fragmented and do not offer clinicians a method of quick and easy communication for sharing/discussing clinical cases and handover information.
To overcome this, medical professionals are using personal mobile devices to support their work, potentially compromising patient privacy and security.
Malnutrition remains a significant health issue for Australian hospitals affecting approximately 35-43% of patients. Evidence indicates malnutrition is associated with adverse outcomes such as prolonged length of hospital stay and frequent readmissions, increased risk of infection, falls, pressure ulcers, increased health-care costs and mortality.
Food waste within the hospital environment is another ongoing challenge for organisations, which has both environmental and economic impacts. Monitoring food waste to address this challenge is difficult due to the of resource intensive nature of collecting timely data of suitable quality and quantity.
Magnetic Resonance Imaging (MRI) scans are used by clinicians to diagnose and treat many common neurodegenerative diseases and disorders including Alzheimer’s dementia. To measure the progress of these, and to understand healthy brain functioning, brain images can be ‘segmented’ and visualized using many software packages and tools in Linux with specific dependencies.
Brain image segmentations can also be used to track tumor location and growth automatically, and take a lot of the manual labor out of identifying abnormalities in the brain. Unfortunately, these tools often take up to 24 hours on a standard computer to produce these segmentations, and require specific expertise.
Recently, ML algorithms have been able to segment brain images orders of magnitude faster than other methods, though the available ML models and packages are largely unusable for clinicians. Clinicians and researchers often do not have the time or expertise to process data in this way. These brain segmentations are an incredibly useful tool for clinicians to make judgements about disease progression and diagnosis. We would like to make using ML easy in the clinic.
We would like to implement a user-friendly cross-platform interface that reads-in MRI (DICOM) data from a clinic or research institution, and provides accurate brain segmentations using ML that can be used as an assistive tool for diagnosis and tracking of disease and disorder. We will call this “Machine Learning for Medical Imaging”, MLMedic, which could automatically detect tumors and growths and measure their size and shape. We would like to design a GUI that applies pre-trained ML models on data collected in the clinic for an easy ML interface for the masses.
About 50% of doctors experience burnout - a trifecta of symptoms which include emotional exhaustion, depersonalisation, and reduced professional efficacy. This has negative effects on their personal wellbeing which contribute to increased direct and indirect costs to health services, and on the quality of care that can be offered to patients. The factors contributing to burnout in the medical profession are multifactorial but one of the biggest problems is that the historical culture within the profession maintains and normalizes burnout. These include, but are not limited to: chronic and high levels of stress; long working hours; irregular working hours; increasing litigiousness; archaic and outmoded hierarchical structures within the profession; arduous exams on top of difficult working conditions; extrinsic and intrinsic expectations of doctors which are difficult to meet; and a culture in which seeking help is perceived as a weakness, a diagnosis of a mental illness is deemed a threat to one’s career, and inadequate provision of individual and systemic interventions to help at-risk doctors. These factors, and others, contribute to a profession which has the highest rate of suicide of any profession (Medscape, 2018).
Whenever humans or animals are involved with a research study, investigators must first satisfy the wider community that the benefits of the research outweighs potential risk to the research subjects. To fulfil this obligation, investigators undertaking new research studies must submit an application to the appropriate institutional ethics committee beforehand. This application outlines the protocol that will be followed during the course of the study, and the steps taken to minimise harm to research subjects.
Once ethical approval is granted, it falls to its investigators to maintain all approvals and permits while they conduct their research study. Modifications to the research protocol must first be approved by the ethics committee before they can be implemented, and the investigators must regularly report on the outcomes of the research study to the approving institution. These requirements generates large amounts of paperwork for the investigators, which takes a substantial amount of time away from research. The ad hoc systems many researchers create to manage their ethical approvals can be difficult to manage efficiently, and misplaced documents can lead to delays in future ethical approvals, or even the termination of a research project. What we need is a rigorous system that any laboratory can readily employ for easily managing these important documents.
Osteoarthritis and tendinopathies are painful conditions that affect more than 30% of Australians, resulting in poor quality of life. Despite decades of research, conservative treatments for these diseases are still limited to drugs to manage pain and no other cure exist. More recently, exercise programs to strengthen muscles have shown promising results in reducing pain, but outcomes vary across people and stages of disease, making a generic one-size-fits-all approach ineffective. What is required is a novel and personalised approach to promote tissue regeneration within the human body.
Musculoskeletal tissues, such as cartilage, tendons, and bones, are repetitively loaded during activities of daily living. Optimal forces acting on tissues help maintaining tissue health. However, when tissues are subject to altered forces, then this balance is disrupted, leading to several debilitating and degenerative pathologies such as osteoarthritis and tendinopathy. Restoring optimal loading can regenerate damaged tissue, but it is impossible to directly measure what happens inside a person’s body because placing sensors inside someone’s tendon or knee could be very unpleasant (and unethical). That is why we have created digital twins that can predict what happens inside a person only by using external sensors and math. We are now able, via digital twins, to calculate all the biomechanical variables that affect tissue health in an individual, but these need to be visualised in real-time, so they can be instantaneously changed to bring them back to optimal level.
Currently, we would like a way to have real-time visualisations of digital twin data- We have a mechanism for creating the data in real time but have no way to visualise it yet.
banner image courtesy of Dr Nick Hamilton