UK Research and Innovation | Survey

Good idea
1st Submission Gold Trophy

DigHealth4Africa

An Artificial Intelligence (AI) enabled digital platform for the management of infectious diseases in Africa.

One of the biggest challenges affecting the health systems in Africa is the lack of standardised mechanisms for patient’s information management, lack of enough experienced personnel and lack of interconnected communication among the key players in the health sector. Take an idea:

-          Patient x goes to clinic A, clinic B and hospital C in the same month and given different drug prescriptions, but there is no any track of the patient’s drug history.

-          A pathologist or a radiologist has to manually analyse slides/images of over 10 patients per day. This is autonomous, erroneous and time-consuming.

-          The governments send drugs to drug authorities which then send the drugs to the health centres, but there is no established communication channel for transparency and prediction of which drugs should be stocked in the near future. Furthermore, external funders usually based on some of the decision to fund the health sector on political influence, not statistical facts as there is not enough evidence.

How could digital technologies provide a solution to this challenge?

The proposed Dig4Health will integrate patient information management, expert system picture archiving & communication system, a public data repository for data science research, digital payments.

Gone are the days when internet coverage was poor when access to phones was limited when hospitals had their own patient management systems. Every journey begins with a few footsteps. With AI and machine learning, we can revolutionize the health sector in Africa for sustainable health and wellbeing.

edited on 4th December 2018, 18:12 by wasswa william

Peju Adeniran 2 months ago

AIs will work with diagnostic and treatment algorithms to make standard protocols across disease states, for more efficient patient management.
It would, if gotten right, reduce the gaps identified in human resources for health in Sub Saharan Africa.

Reply 1

wasswa william 2 months ago

Thanks Peju. You are very right. AI can play a significant role in reducing some of our African Health Challenges

Reply 0

BHUVANA K.B 2 months ago

Great idea. Patient information management is equally important for non communicable diseases or any chronic disease management. Especially also important for patients with multimorbidity. Here in India as well, we have increased fragmented care. Disease drug, drug drug interactions could be avoided through electronic health records linked across the levels of care. (primary/ secondary/ tertiary, government/private sectors)

Key stakeholders should come together including IT departments along with the professionals, patients and caregivers.

Reply 0

wasswa william 2 months ago

Thanks, Bhuvana. Indeed such an integrated transparent platform can be very good. Luckily many countries in Africa now have national Identification numbers (IDs) which are currently being used to register voters. We can tap this into health systems for efficient monitoring and personal health.

Reply 0

Charles Clerck 2 months ago

Status label added: Good idea

Reply 0

wasswa william 2 months ago

Thanks Clerck. Together we can change the health status in Africa

Reply 0

Luigi Sedda 2 months ago

Good point William. And I will go further than that by suggesting a pan-african digital initiative that can provide surveillance as well as control of infectious diseases; but also good denominators for estimation of risk and spread.

Reply 0

wasswa william 2 months ago

You are right Luigi. That is really a good integration

Reply 0

Kat Valero 2 months ago

Hi all,
My lab is currently interested in developing such digital / AI-based health solutions in the context of eco-environmental forecasting. We have developed collaboration with the Universities of Lagos and Ibadan but are looking to network further. Cheers, Kat

Reply 0

Osu 2 months ago

Hi Kat, my group will be interested as well. Anyways to connect with you? And what do you require?

Reply 0

Kat Valero 1 month ago

Hello Osu, you can write me an email to k.wollenberg-valero@hull.ac.uk I'm happy to discuss

Reply 0

View all replies (2)

Carol Crean 2 months ago

Good idea.
We have been working on textile electrodes for wearable sensing and have had good success with a sensor for lithium - which is prescribed to treat bipolar disorder. According to a collaborator who is a clinician and works in Ethiopia, it is sometimes prescribed at ineffective doses, and monitoring is not always possible.
We are trying to secure funding to validate the sensor response in patient samples and have UK and EU-based clinicians interested. Once validated it might prove useful in remote regions where results could be sent digitally to your Dig4Health information systems.

Reply 0

BHUVANA K.B 2 months ago

Hi Dr Carol. I am Dr Bhuvana, Pharmacologist. Good to know about sensor for lithium. Drug which has got narrow therapeutic range, need to be cautious before increasing the dose of the drug. Noncompliance could cause therapeutic failure. Exceeding the range causes cardiovascular toxicity, can lead to coma, convulsions and death. By estimating lithium levels, by therapeutic drug monitoring would be combersom. Do you estimate levels for any other drugs other than lithium?

Reply 0

Carol Crean 2 months ago

Hi Dr Bhuvana, we are working towards other analytes but do not have anything to report yet since this is technically challenging.
Our intention is to use non-invasive techniques to access the lithium in interstitial fluid. We chose lithium because of the narrow therapeutic range.

Reply 0

BHUVANA K.B 2 months ago

Sure Dr Carol. Yes I understand that it's non invasive.

Reply 0

View all replies (2)

David Stern 2 months ago

William I like the basis of your idea a lot but there are a number of notoriously difficult problems with implementing a system like this. I have some experience in developing software solutions and would probably turn down such a task if it was presented to my team because of potentially insurmountable challenges in creating a successful implementation. I give you three challenges which would come to mind and would really like to get your thoughts on how they could be surmounted.

Data sharing: There are serious privacy, ethical and financial constraints that could lead to clinics and health centres having objections to sharing their data in this way. In many cases these are the primary constraints to building such a system. Have you thought about how you would both incentive the sharing and ensure that valid privacy and ethical concerns are met?

Financial model: If such a platform existed what do you see as it's financial model? It could be created as commercial, open, governmental, multinational, charitable, ... each would have it's own advantages and challenges. What would you currently be thinking?

Responsibility: If such a platform existed and did provide advice to health care professionals based on AI, who is legally responsible for any negative consequences of that advice?

Reply 1

Josephine Muchogu 1 month ago

Hi Wasswa, I think that it is a fantastic, rational idea. However, I have to agree with David with regards to the ethical challenge posed by the right to privacy and confidentiality, and research consent.
Clear structures must be set in place for patients' consent on data usage.
Stringent measures must also be put in place for access to the AI system.
I look forward to seeing how this WILL work out.

Reply 0

wasswa william 1 month ago

Thanks Josephine, you are right. Clear structures need to be in place for such an integration.I

Reply 0

wasswa william 1 month ago

Thanks David for your suggestions. I truly agree with you especially in terms of data sharing. On top of teaching as Biomedical Engineer at University, I am also a software developer and a director of an ICT firm in Uganda purposely empowering SMEs to use ICT to cope in this digital era. We have implemented hospital management systems in around 4 private hospitals but from my experience I agree with you that an integrated patient management system or personal health system may really be very difficult to achieve unless strong ethical measures are put in place. However this can easily start with the public hospitals.
I also note that many countries like Uganda have National IDs with chip, but these IDs are not fully utilized. How can our governments tap into such devices for proper management of patients. Data sharing can have access levels, may be hospital x can only access your last dose and where you got it from, not necessarily all your history??

Furthermore on data sharing, in Uganda you can find a pap-smear moved from cancer screening unit x to cancer screening unit y where there is an experienced pathologist to diagnose such a complex pap-smear. Why NOT have an integrated system for all cancer units where such images can be viewed instantaneously by different experts at the same time.

Lastly still on data sharing, there is lack of clinical data for research in Africa. I have worked on a tool for automated diagnosis of cervical cancer from pap-smear images, but access to public pap-smears in Africa population was impossible until I had to prepare my own dataset for final evaluation. Furthermore development of prediction models for a number of diseases require patient data. How can we help the scientific community have such data to advance research in Africa?

Reply 0

View all replies (3)

Davis Ntwiga 1 month ago

Hi Wasswa. Dig4Health can incorporate an assistance to the patients to identify genuine and fake drugs which is a menace in some African countries.

Reply 0

wasswa william 1 month ago

True Davis including prediction of where it is more needed.
It is very common to find that drugs are destroyed in medical stores because they have expired yet health center x needed these drugs at some time, t.
This always puzzles me with a question..
How can we improve the Quantification and Forecasting of required essential Drugs based on updated guidance and history data??????

Reply 0

Osu 1 month ago

So this happens in Uganda as well?

Reply 0

BHUVANA K.B 1 month ago

Wasswa, does drug utilization studies answer your questions?

Reply 0

wasswa william 1 month ago

Thanks Bhuvana. I think from a policy based approach they may help.
This is because for example during immunization, most immunization centers do not have exact data of how many parents will turn up on the day of immunization, hence some times immunization drugs are not enough in some remote centers. There is also a tendency of many mothers forgetting the return dates...

Reply 0

BHUVANA K.B 1 month ago

Actually, now Pharmacovigilance is a mandate for immunization as well. So it's essential to follow up patients on adverse reactions with immunizations and it's documentation.

Reply 0

BHUVANA K.B 1 month ago

Genuine and fake drugs differentiation should occur much before reaching patients. There should be good quality control systems to conduct Bioavailability and bioequivalance studies. If generics are manufactured. So Good manufacturing standards should be followed.

Reply 0

View all replies (6)

BHUVANA K.B 1 month ago

Would not mother / parent get reminders? Are not they followed up, once registered? For routine national immunization schedules

Reply 0

Kat Valero 1 month ago

There are existing resources that can be used for such purposes to build up a digital health infrastructure; for example via our collaborators working in the nexus of the (Wellcome funded) https://h3abionet.org/

Reply 0

wasswa william 1 month ago

Thanks @Kat for that link to H3ABionet.
Actually, our DigHealth4Africa is a very integrated framework with different modules, but the one we want to start with at Mbarara University of Science and Technology is the DigPath, an integrated Platform for cervical cancer management. I have worked on it for my PhD and now trying to see how we can implement some parts. It has 4 modules.
1. A low cost microscope slide scanner to produce fast high quality pap smear images
2. An automated pap-smear image analysis tool
3. An Expert system for cervical cancer management (patients information management including sending reminders to patients, connecting the different cervical cancer screening units so different experts can work on the same pap-smear virtually at the same time and send complicated pap-smears to experts, cervical cancer recurrence/occurrence prediction, automated risk factors assessment/evaluation, tumor board meetings management)
4. An open repository. (this will provide labelled data (without patients information) to the public for research (pap-smear images, risk factors evaluation)

Reply 0

Kat Valero 1 month ago

Hi Wasswa, the approach you outline is not un-similar to the one we propose in a (pending) GCRF application, which however has a different health related focus than the one you mention. Let's keep pushing for digital health infrastructure development and support these endeavours in the future.

Reply 0

wasswa william 1 month ago

True @Kat. We just need to keep in touch and looking for possible funds to implement such endevours in the future.

Reply 0

View all replies (2)

Davis Ntwiga 1 month ago

@Wasswa. To make the application even more comprehensive, it can track sections or zones in a country where specific diseases are a menace.

Reply 0

wasswa william 1 month ago

True Davis

Reply 0

Mohamed Ajuang 2 weeks ago

Very intuitive idea, and have you thought about adding Space time boxes and the element of geography (GIS)?

Users tagged:

Reply 0

wasswa william 2 weeks ago

Actually that is a good idea for more relibale data.

Reply 0

Mohamed Ajuang 2 weeks ago

Food for thought when you will be implementing the project

Reply 0

View all replies (2)

Nervo Verdezoto 2 weeks ago

It is a nice idea, however I still wonder about the multiple conflicting perspectives from different stakeholders and how you would gather and share their knowledge? Rather than taking a top-down approach and thinking that AI will solve the problems, a bottom-up approach will be also needed to unpack the everyday challenges with the situated infrastructure before, during and after the deployment of the technology. Technology would not only change the work practices but also might disrupt different procedures that are currently in place. The ethical considerations should also be taken into account throughout the whole duration of the project. A good idea could be to invite different stakeholders to a co-design workshop and explore with them the potential benefits and challenges of using AI in the care setting.

BTW, we are having a workshop in Glasgow this year on "Unpacking the Infrastructuring Work of Patients and Caregivers around the World" at CHI 2019, Glasgow, UK, May 4th 2019. https://chi2019healthworkshop.weebly.com/

Maybe any of you could consider to apply and share the infrastructural challenges in the African context.

Reply 0

wasswa william 2 weeks ago

Thanks very much for your comment @Nervo. But for as far as we have gone, we have really been working with the different cancer institutes in Uganda using the Design Thinking Approach. We have put stakeholders at the center of the design process. We are working on workflows for the current protocols to ensure that AI does not gretaly change these workflows unless otherwise.
For the conference, let me see if one of us can attend. I am currently in Glasgow at University of Strathclyde until March 2019.

Reply 1

Nervo Verdezoto 2 weeks ago

That is good then as nothing was said about patients and health professionals involvement in the proposal before. Try to look for previous experiences (e.g., unintended consequences, users, usages and workarounds) of deployment of systems in healthcare settings, that could inform your project activities and the design of the tech.

Looking forward to hear more about your project and any pilot data that you have collected so far.

Reply 0

View all replies (2)

Teresa Kinyari 2 weeks ago

This is a great idea. Would the use of unique identifiers mitigate on the risk of breaching patient confidentiality. How would you obtain consent for the use of these data?

Reply 0

wasswa william 2 weeks ago

Thanks @Teresa, yah the concept of unique idetifiers is a good one for anonymity. The government recently just approved a policy about health data sharing. There will be a possibility of providing some anonymous data for research purposes especially the pap-smears and history information (longitudinal register)

Reply 0

lauren clarke 2 weeks ago

Status label added: 1st Submission Gold Trophy

Reply 0

wasswa william 1 week ago

Hi Team, I was so surprised today, to realise that some complicated pap-smear slides are transported from one cervical cancer screening unit to another unit which is about 270 km where there are expert cytopatholgists. This is time-consuming and expensive in the long run. With the implementation of our proposed DigPath Platform, all major cervical cancer screening centres will be connected and different experts can view a slide virtually and emit diagnosis without the need to physically transport such slides.

Reply 0

Share