XPRIZE Healthcare Everywhere

Less than half the world population is covered by essential health services. Roughly 800 million people spend at least 10 percent of their household budgets on health expenses for themselves or a family member. Developing countries and even remote areas in nominally wealthy nations lack medical funding, infrastructure, and skills.

What if we could crowdsource diagnosis and treatment? Analyze and compare insights, and synthesize them into a recommended course-of-action that can be performed even in remote areas with affordable medical kits?

This prize would challenge teams to deliver high-quality medical care in a cost-effective way to far-flung locations.

Potential metrics would include:

  • Accessibility
  • Affordability
  • Ease of use
  • Upskilling of frontline health workers
  • Possibility of continued care

Solutions could integrate everything from crowdsourcing expertise to AI diagnosis to drones that deliver medications to robots that perform surgeries.

Do you think this would make for an audacious prize competition? What exactly would the winning team need to do, and what would be the likely impact of this prize?

@addy_kulkarni, @Anupama_Reboot, @rajpanda, @scveena, @Fatima, @JoanneP, I hope you can advise us on this idea for an XPRIZE. What are some of the solutions you think we should incentivize?

2 Likes

@Mellie64, you may have thoughts on this idea for an XPRIZE competition from the perspective of rural healthcare. Does this sound like the sort of solution that could help you?

@tingatclick, your point of view would be valuable to us here as well.

Thanks!

1 Like

We have several potential winning-team-will statements for this prize idea:

  • Converge various existing diagnosis and therapeutic tools.
  • Develop a network of interoperable AI subsystems, whereby an “executive” level AI subsystem can call upon multiple specialised AI subsystems to provide comprehensive analysis, good accuracy in its predictions and recommendations for optimum outcomes.
  • Nominate a specific medical/therapeutic target that obeys basic criteria of “public good” relevance, and design an AI appropriate for that target.
  • Develop a clinical tool that synthesizes relevant medical research in collaboration with clinicians, analyzes electronic medical records, and makes hypotheses and/or recommendations to caregivers.

@caitrin, @farahelsiss, @mitjal, which do you prefer?

@MachineGenes and @ymedan, you may also want to weigh in on this, as the prize sketch incorporates part of the feedback you provided in the “Dr AI” discussion.

1 Like

Hey @NickOttens , for clarity this is a separate prize from Dr AI?

Even here in Australia we have these problems, in what are called “rural and remote” locations. The key problems are access to medical sensors/devices and appropriate specialist expertise. This prize is presumably concentrating on the latter.

  • Essential for any such architecture would be fully-explainable AI: the AI needs to be able to communicate with a local doctor, paramedic or if necessary, simply the patients themselves, communicating risks and options for informed consent. It also needs to be interactive, so it modifies its behaviour/therapy suggestions (including specific doses) based on the responses of the responsible human on the ground. Without that, this is a non-starter: it would be completely unethical to deploy such a technology.

  • The notion of interoperable AI is very difficult; again unless you have fully-explainable AI (which indeed is possible-- we’re about to demonstrate an example in medicine-- but probably not if using neural networks), coordination of clinically-appropriate interoperable systems is science fiction. (It’s the sort of thing everyone thinks is a good idea that ‘‘should’’ exist, but building it is another matter. This is where the Dr AI specification made much more sense from a practical perspective.)

  • What does ‘‘comprehensive analysis’’ actually mean? Again these are terms that sound good, but it’s difficult enough to prove with a human doctor, let alone an AI.

Sorry to be difficult, but the Doctor AI specification appears feasible, whereas aspects of this one appear to me to be a bridge too far.

The medical solutions for last mile shall be scalable and sustainable at that scale. Also the behavior change aspect of healthcare professionals shall not be underestimated. Especially as last mile public health programs go hand in hand with local governments, system-fit innovations will have less hurdles for implementation. Can we consider these points while deciding the metrics?

Thank you both for your comments, @MachineGenes and @addy_kulkarni!

The “Dr AI” aspect has informed this prize design as well as the Health-Optimized Home and Digital Twin. But we’re still working on figuring our whether it should indeed be a criterion in all three prizes, and if so how. So your feedback is much appreciated!

Would it make a significant difference if the solution is for use by healthcare workers (possibly with minimal training) or patients?

Would we even need an AI to diagnose common ailments, like diabetes, high blood pressure, malaria, and pneumonia, and prescribe the corresponding treatment?

Hello @NickOttens, yes there’s a big difference between building a solution to advise paramedics/healthcare workers versus one for patients themselves, due to regulatory, safety and user interface issues.

Diagnosing an ailment and prescribing treatment are two very different tasks. Prescription predicated on an existing human diagnosis is lower-risk, but surely not available for the remote context here. (By the way, you can’t “crowdsource” diagnosis and treatment. I think it’d be a non-starter from an ethical perspective, as well as simply not working very well.)

As regards prescribing the corresponding treatment for common ailments, yes, sometimes you do need AI (or some other form of sophisticated intervention). To put this in context, two examples:

  1. Only 25% of people with type-1 diabetes (T1D) actually manage their insulin appropriately to achieve adequate blood glucose control. The remaining 75% fail, risking hypo- and hyperglycemia, with serious medical consequences. Poor blood glucose control in T1D costs the US economy over $22.5bn each year in medical costs and lost income.

  2. Consider flu-like symptoms. In the United States, Europe or Australia, the most likely diagnosis would be flu (pre-COVID-19). In Papua New Guinea, it means you have malaria. The treatment for malaria is heavily dependent on geography: if I’m at risk of getting it in Papua New Guinea the appropriate intervention is different from if I’m in Thailand or Rwanda. Context is everything.

We’re renaming this one “Healthcare Everywhere” and not requiring an AI component, leaving that up to competing teams.

The goals of the prize would be to:

  1. Provide diagnostics and basic, or appropriate, medical care to everyone, everywhere.
  2. Enable users in low-income countries and underserved communities to gain basic health provider skills.

The winning team would built a portable hospital “kit”, adaptable to local contexts, and with built-in privacy safeguards, that packages a comprehensive suite of advanced telemedicine tools, rapid diagnostics, and first-line treatments that can be used by anyone, anywhere.

Testing and judging criteria could include:

  • Must be so intuitive that everyone? or a person with basic skills? can use it.
  • Low-cost. But how low-cost?
  • Must be portable and rapidly deployable, including to areas with no or low-band internet.
  • Adaptable to local context (needs and language).
  • Use easily sourced technology (for example, cell phones).
  • Interactive.
  • Enable continued care.
  • Scalable (if the competition allows submissions for specific diseases or medical conditions).
  • Minimal maintenance and repair required.
  • Future-proof (so future data and tech can be plugged in).
  • Connected to different relevant systems/caregivers.
  • Upskills local healthcare workers.
  • Non-invasive, to avoid drawing blood or collection of urine samples.
  • Open to crowdsourcing elements from non-experts (like contextualization?)
  • Incorporate scientists and (mental) health workers from low- and middle-income countries.

You can see some of the outstanding here. Please feel free to pick this apart and comment on all bullet points!

The regulatory question is also highly relevant. Should we focus on solutions that don’t require regulatory approval? Or would that limit us too much?

2 Likes

Another outstanding question is what level of care, or how much care, the tool should provide. When we say it should package a “comprehensive suite” of advanced telemedicine tools, rapid diagnostics, and treatment, what exactly does that entail? How can we phrase this in a way that solution-agnostic but also challenges teams to be audacious and built solutions way ahead of what is currently on the market?

1 Like

I would recommend you have a look at how the worlds armed forces have their field hospitals & clinic systems set up for these " Health Care Everywhere" modular systems. They do a pretty good job and they are not just for trauma if you have not already looked into that.

@MachineGenes, @addy_kulkarni, @Lee, thank you for your feedback on this prize idea!

The advice we got from our founder and chairman, Peter Diamandis, is that this prize, as currently written, is too broad. He suggests we should make it narrower and more concrete.

His suggestions included:

  • Test for infectious diseases.
  • Test and diagnose top X diseases.
  • Support pregnancy/natal care.

What is your take? When thinking about healthcare needs in developing countries and rural areas, how could we make the biggest impact?

@pglass, @shamakarkal, I know you previously advised on our Frontline Health Prize Design. You may have thoughts on this idea for a prize competition as well. Please join the discussion!

1 Like

I also commented in the Frontline health prize design. I agree Healthcare Everywhere is a good idea, but needs a focus so that it becomes feasible because improving healthcare everywhere is extremely difficult. The focus I like to see is not in dealing with X number of diseases, rather building the system that can deliver some care improvement into rural and underserved regions initially and can be expanded to deliver more and more care continuously in the same system. Such system-level innovation will be more productive than disease-focused approach. I think, “Healthcare Everywhere” prize can put in a reusable, extensible and scalable system in place first, from which more innovations and improvements can be enabled on an ongoing basis.

2 Likes

Thank you for the feedback, @ajchenx! That sounds intriguing, could you elaborate a bit for me? What exactly would we challenge teams competing in such a prize to build?

You may challenge them to build a system connecting rural clinics and city teaching hospitals on which they can deliver high quality disease screening, clinical prevention, doctor referral, disease early detection, real-time clinical collaboration, patient education, etc. to rural doctors.

Several key requirements include:

  1. re-usable system: It can gradually deliver services for more and more diseases. we can’t repeatedly build different dissemination systems because it’s too expansive. [cost-effective system for assisting most aspects of primary care]
  2. highly automated system: for example, AI-based cancer risk prediction capability on the system for rural doctors to use daily, and at the same time, teaching hospital can continuously improve the AI model from data without sending doctors to rural clinics. [no big need for doctor-resources from city hospitals]
  3. driving force from city hospitals: City hospitals must have incentives to connect with the rural clinics and drive the services on the digital system. Without hospitals’ driving the “healthcare everywhere”, it is hard to sustain. [sustainable network]

Such system will be difficult to build, but it will have much higher chance to succeed compared to all the approaches we have seen so far. It’s a big challenge. Someone may come up easier ways to achieve it.

Given that the At-Home Testing prize will probably be a blood test, we don’t want to focus this on infectious diseases. Top X diseases in a region might also be a little superfluous, given that some of the top killers in Sub-Saharan Africa and South Asia can be detected with a blood test.

We have alternatives:

  1. Cheap, portable MRIs. The most innovative product I could find it is the mobile MR imaging system sold by Hyperfine for $275,000. If we can make something much smaller and much cheaper, it could be a huge breakthrough. Although the challenges there are a) interpreting the results for diagnoses - who’s going to do it? and b) what’s “cheap”? @RahulJindal points out that even a $1,000 device would be unaffordable in remote parts of low- and middle-income countries.
  2. Addressing unmet surgical needs. There are models for cheap and effective eye surgery. Are there other unmet surgical needs you would recommend we focus on? What would be the most impactful?
1 Like

I have visited some well-known models of cheap and very successful eye care in India. So why duplicate what is already known. If you wish to concentrate on eye surgery, perhaps, replicate one of the models into even cheaper version.

1 Like

Another issue with a ‘very cheap’ MRI would be overuse, over diagnosis and misdiagnosis. This would result in anxiety for the patients and their families. An example is the controversy about the age for mammograms. The guidelines for mammography have changed frequently and there is lack of agreement between the radiological societies of different countries.

My group from India has shown that ‘unmet surgical need’ varies from country to country and even within a vast country such as India. Even within a big metropolitan city, unmet surgical needs can vary between different income levels. Our work in a slum of a metropolitan city showed that the unmet needs were: extremity injuries/wounds, kidney stones, cataracts, abdominal pain including external hernias, and menstrual disorders (heavy menstruation). See - https://link.springer.com/article/10.1007%2Fs00268-020-05502-5

My group is doing a similar study in a high-income and mid-income area of the same city and the unpublished results show a significant variation of the surgical unmet need.

So to summarize, the first part of the competition would be for a group to define and study the surgical unmet need and then devise solutions for those unmet needs.

Thanks for your comments, @RahulJindal!

Looks like eye surgery doesn’t need an XPRIZE, based on the links you’ve shared.

I’m skeptical of a competition that allow teams to pick different unmet surgical needs. Then it will be difficult for judges to compare their solutions. Usually the competition defines the problem.

Ideally we would select one problem (for example, untreated kidney stones) where there isn’t a lot of innovation, but where a relatively low-cost solution could make a big impact on the lives of many people.

1 Like