What do the top ten most commonly prescribed medications, including treatments for hyperthyroidism, asthma, heart disease, and ADHD all have in common? They all target membrane proteins. In fact, despite membrane proteins making up only a third of proteins in the human body, over half of all medications target them including treatments currently in development for COVID-19. However, the structural and functional mechanisms of many of these proteins remains a mystery making the effective development of new medicines highly wasteful and difficult.

In 2017 the Nobel Prize in chemistry was awarded to Jacques Dubochet, Joachim Frank and Richard Henderson for their contributions to the development of cryo-electron microscopy, a technique that has led to a revolution in finding structures of membrane proteins by allowing researchers to capture 2D images of individual molecules and use these to create high resolution 3D models. These high resolution 3D models can then be used as the basis for a computational drug discovery platform. But despite these advances, building a high quality model of a membrane protein is not a trivial task.

All datasets have anomalies, in cryo-electron microscopy data these come in the form of broken particles. One key challenge in the analysis workflow is sorting high quality 2D images from images of broken particles or other false positives. Current platforms use a maximum likelihood approach to 2D classification that groups similar looking images into classes that can then be manually determined to be good, and used in further analysis, or bad and removed from the dataset. While useful on a very large scale, this method takes a significant amount of time and worse, it never truly removes all the bad images and just a few bad images in a dataset of thousands of particles can significantly hamper high quality 3D reconstruction.

This is where ‘CryoFilter’ will come in. We envisage a machine learning based software tool that can automatically detect and remove bad particles from the datasets, leaving only the good particles to rapidly create the 3D high resolution structure of the membrane protein. CryoFilter will integrate seamlessly into existing workflows while removing the crucial step of manual sorting. Once in use, CryoFilter will save biomedical researchers in hundreds of groups around the world weeks to months of time which they now spend manually selecting images, significantly speeding up the discovery time.

Auslan (Australian Sign Language) signs could be recognised by a machine learning model, which could enable sign-based control of technology (e.g. “Siri for sign language”), and automated interpretation/translation for electronic media. Machine learning approaches require large datasets to train, and such a dataset is not available for Auslan. One way of creating a dataset is by crowdsourcing. If there was a community portal available, expert and learner signers could contribute videos of their signing to the dataset, along with appropriate metadata about translations, their level of Auslan proficiency, and who they want to have access to their videos. Such a community portal could also become a useful tool for learner signers, if open-access videos were made available for viewing. I propose a HealthHack team could create a prototype for such a portal. The ideal for this would be a system which enables:

1. fluent/Deaf signers to create videos of Auslan signs, phrases and sentences;

2. learners of Auslan to view sign, phrase and sentence-level videos of Auslan produced by fluent/Deaf signers and advanced learners;

3. learners of Auslan to create videos of their own Auslan signing, at sign, phrase and sentence levels, prompted by pre-existing videos or written prompts;

4. creators of videos to report metadata, such as Auslan dialect, their Auslan proficiency, and the visibility of their videos (i.e. for use by researchers only, available to the community, or available to anyone);

5. community members to downvote or report videos which are unclear or incorrect.

This portal could then be used to support a community of Auslan learners to communicate, practice, and learn from fluent/Deaf signers and more experienced learners, increasing their vocabulary and learning correct Auslan grammar, while also creating resources for use in machine learning training.

Currently, there are limited systems in place in the public health system that cater to people with sight impairments. For example, a patient attended the public dental clinic and was upset to be filling out medical history forms. The clinicians found out much later that they suffered from dyslexia, which prevented them from accurately filling out the forms. The medical history was then taken verbally but the appointment ended up running late. Overall, the current systems are ineffective in maintaining patient autonomy for those blind or visually impaired when filling out forms and medical history crucial to their appointments and treatment plans. Often the responsibility on support is delegated to available receptionists who are often untrained and unprepared for this. Although technology is currently in place such as speech to text recognition software or medical terminology databases, these have not been utilised to assist with the taking of medical history for vision impaired peoples. This in effect, makes them reliant on the current system which is flawed and often discourages patient attendance and compliance. We hope over the weekend to develop an app with some basic medical forms (templates will be provided) that may be filled out by speech to text or have options for selecting text size and changing colour to accommodate a range of visual impairments.

More and more young people are being diagnosed with inflammatory bowel disease (IBD) and lack access to multi-disciplinary IBD care. Crohn’s and Colitis Australia (CCA) have shared research findings from a national patient experience survey of IBD care. The 2018 research report demonstrates there is a high burden of Crohn’s disease and ulcerative colitis in Australia. Australia has one of the highest rates of prevalence and incidence in the world and each year more and more young adults are diagnosed. Over 85000 Australians have Crohn’s disease or Ulcerative Colitis and more than 5 million world wide have the disease. Crohn’s disease is incurable and associated with a 47% increase in the mortality risk. Not much is known about IBD in the general public and sufferers are burdened with a hidden struggle that affects all aspects of their life. These patients are commonly young, unwell with chronic disease and have struggled with the condition for more than a decade. The reality is that this disease affects relationships, mental health, social interactions and employability.

Oscar the mind-gut guru is a digitally based support system that looks at disease holistically exploring ways that young adult sufferers can manage their experience of IBD. Oscar communicates to young people, by offering mindfulness tools to transform negative emotions that IBD brings. Oscar ( a web based application) will enable young people to connect, support and demystify the experience of living with IBD. With the help of Health Hack 2020, Oscar can help assist CCA to drive improvements in quality of care and services for Australians living with IBD. If we can look at the disease holistically we can identify all areas of improvement that are currently costing the healthcare systems financial burden. It is estimated that the total cost of caring for Australians with IBD is estimated around $3.1 billion each year, and that cost is expected to rise as the impact of the disease becomes more understood. Oscar not only has the potential to save lives, but also the financial burden of a chronic illness.

Contact tracing is a very useful strategy to control the COVID-19 pandemic. Around the world, there has been a worldwide use of contact tracing apps to achieve this. At the moment, in developing countries, it has been a struggle to encourage people to use contact tracing apps since they do not feel confident with their data being stored by their governments, and they fear their data will not be used properly.

I consider the reason for the failure of the adoption of these apps is the way the solution was presented. These are apps made by governments, who ask upfront for far too much information about the user. I believe the strategy should be changed. People like to feel they are in charge of their digital personas, and to have certain anonymity online that they can control. People already give too much information about their lives to internet giants like Facebook and Google, for example.

I envision a solution where people can create a ‘digital identity’ of themselves and behave more like on a game or social network, than a contact tracing app. I also envision that this contact tracing app, through gamification techniques, encourages users to keep social distancing and to follow self-care measures, and where users do not give their data, unless it is absolutely necessary (a confirmed covid-19 case or he/she has been in contact with one).

At HealthHack, I would like to build a minimum viable product that we could use across the developing world. Many majors in developing countries like where I come from (Colombia) would be interested in a solution like this, if it is available.

People from marginalised groups can sometimes wonder if they are the only one who feels like they are being treated differently. It can be cathartic when they realise that they aren’t the only one feeling this.

Both of these situations can make people from marginalised racial groups feel helpless which can lead to increased risk of health problems.

This solution will capture, curate and share stories and data about racism. This could also be used for capturing information for other marginalised groups.

The ideal solution would have three areas: Capturing the data easily into a database, Curating the data in a crowd-sourced way using web technologies, and Sharing that data to the world in an easy to use web interface.

Options for capturing the data might include:
1. Using a sharing app that can submit directly to the database
2. Create a twitter hashtag and read it in similar to Thread Reader App.
3. Or subscribe to a particular Twitter feed.
4. Extensions for Chrome and Firefox

Options for curating the data using a web interface could include:
1. Tagging and autocomplete based off previous metadata.
2. Tagging and autocomplete using and ontology (defined dictionary).
3. Adding a description or an editorial
4. Publish / not publish checkbox

Options for Sharing the data:
1. Fast, snappy website that acts like a web-based pocket dictionary
2. Simple and search focused, with some stories on the front (tag and count)
3. Mobile friendly.

List of roles needed could include Front end Software developer, Back-end Software developer, DevOps person, Graphic web designer, end user and data curator.

An example home page is here: https://wireframe.cc/iPjrUQ

As a society we rely on scientifically derived knowledge to make decisions about almost every aspect of our lives including many important matters of life or death. So it’s important that the public in general are able to accurately interpret certain scientific results. Researchers value facts and spend a lot of time and effort making sure that when they publish their work they accurately describe the results they’ve obtained and how they obtained them. They typically communicate their results using scientific language that’s targeted largely to other researchers in their field, however most people are not researchers and most researchers will likely work in an unrelated field. In truth, almost everybody significantly relies on interpretations of scientific results that are passed along by others including friends, colleagues, teachers and the media. If the research has been misrepresented it can have dire consequences.

To make their work more accessible some publications and institutions encourage or require a media release to be submitted alongside the research, however there is currently no efficient way for researchers to predict how that release could potentially be presented and / or interpreted in public spaces and the media (both traditional and social). 

I want to help researchers get better at predicting how their research will be viewed which I hope will improve scientific literacy amongst the general public.

banner image courtesy of Dr Nick Hamilton