Creating 'Contactless' Remote Clinics, using Artificial Intelligence driven chatbots, computer vision and IOT enabled devices

Reposting with permission from ERUDITE Magazine Vol 1 by IAMBSS.

The problem of ‘whom to ask’, ‘where to go’ and ‘what to do next’ is more common than we can imagine. Right now, people (both rural and urban) suffer from these three constraints when one needs some kind of medical advice.

Right now, people (both rural and urban) suffer from these three constraints when one needs some kind of medical advice. The problem of ‘whom to ask’, ‘where to go’ and ‘what to do next’ is more common than we can imagine. The convenient source of medical information is a local chemist or a local quack for most. This delays the care of serious patients who need tertiary level care. The doctor to patient ratio ranges from 1:11,000 to as low as 1:28,000 in some states(1). According to the data published by NCBI 57.3% of allopathic practitioners in India do not have a medical degree and many are not even graduates(2). The government is proposing a solution to give nurses and these ‘untrained health care provider a 1 year “Bridge course” (3) so they can be licensed to give ‘over the counter (OTC)’ drugs, refer patients to the right centre of  care, and become a part of the official workforce. 

These are desperate measures to solve a problem that is already in dire situation. The COVID-19 pandemic just exploded the fragility of healthcare system and now the demand for increasing the “reach” of medical advice as well as it being “contactless” has been realised as a priority in coming years. So, what can be some possible solutions? We should be open to all solutions that can work, be agile in evaluating them, and quick in implementing on ground level. 

Clearly some sort of a system and an organised way of doing things is needed. Telemedicine has become a “Wonder Pill” solution to all these problems in mainstream media. Telemedicine does solve many problems. It does save travel time, the patient’s travel cost, and helps in a contactless approach to see patients. However, certain problems are still not resolved, the major being the number of ‘man-hours’ or ‘consultation-hours’ is still the same. The same deficient doctor-patient ratio that was available offline has now moved online. The number of practicing hours a doctor has cannot change with Telemedicine. In addition, patients still do not know which ‘type’ of specialist they have to go to and how grave their illness is. Some may perceive ‘Telemedicine’ to be a costly affair where they spend money without finding a good solution. And again, poor triaging or the problem of primary care patients lining up for a doctor’s time while causing delay for patients with serious illness remains an unresolved issue. 

To settle these issues a “middle man” or resource of some kind, that can scale and multiply rapidly is definitely needed to “bridge the gap”. Someone who can guide well, easily approachable, reliable, well tested and available for all. Here I propose a solution that can be used to promote the reach of genuine medical advice, and help solve the problems of ‘Access’, ‘Affordability’ and ‘Awareness’ for the rural and urban population of India with regards to the healthcare sector.

Introducing ‘Healthcare Chatbots’ (derived from Chatting Robot)(4), an invention of recent times invented by a very few companies as of today to name a few like ADA health from Germany, Infermedica from Poland, Babylon from UK and ZINI.ai from India (5). 

A chatbot is a virtual robot that is trained in the art of interviewing a patient (history taking). Given some initial complaints by the patient, the bots can autonomously and remotely discuss a case with the patient to figure out what can be the possible diagnosis and what should be their future course of action. Basically the 1st step of the patient care process gets automated (fig 1). And this initial step is of paramount importance which decides almost everything in future on patient care. If done rightly, distraction, misguidance and loss of time / money can be saved. 

In addition, now we have many Internet of Things (IOT) enabled devices that can electronically and remotely, gather data on step 2 and 3 of the process. Devices that can measure pulse rate, blood pressure, temperature, oxygen saturation, respiratory rate, and many other important parameters, including a 12 lead ECG. 

Automating these 2 steps can be easily done remotely by setting up ‘Remote OPDs’ or ‘Satellite OPDs’ by hospitals in their area or region of patient source. Patients need not come directly to a tertiary care center always. This becomes their 1st point of contact in their local area. Infrastructure setup is easy and costs less patients can go here, instead of an untrained healthcare provider, where a trained AI model discusses their case. AI model is able to decipher the possible differential diagnosis, the urgency of the case (triaging), the investigations that may be needed, the ‘specialty’ or the ‘type’ of doctor the patient should be referred.

A local MPHW (Multipurpose health worker) or a nurse can be trained to operate the chatbot as well as devices to collect more information on patient’s vitals. Now as per the triaging and recommendations by the AI bot the whole data packet (consisting of patient’s basic info, patient history and physiological parameters) can then be transferred to the specialist sitting remotely, who on his/her dashboard is able to see the list of patients from various remote OPDs. If a case demands, talking to or seeing the patient, an audio or video call can be made to the local remote OPD. 

Figure 3 A doctor’s Panel showing all details like Basic info, Vitals, History collected by the bot. 

The AI bots and devices along with a local worker can run an OPD, collect data right away, share it with doctors sitting remotely and get the right advice locally. Such remote OPDs can be set up in residential society, university campus, villages, remote islands and other places. 

These virtual Chatbots are not just limited to being stuck in these remote OPDs they can be installed in a website or mobile app as well. For example, a hospital can install the chatbot in their website where people can come, share their ailments and the bot can run an interview online and generate a guidance report as to what could be the cause, what specialty they should consult and how urgent the situation is. Basically, everyone gets the good guidance right away. The chatbots are also able to answer some common questions that most patients have during follow ups or pacify their queries about a procedure or disease by providing them information and answers from reliable sources. Some new bots are even able to run ‘Computer vision’ techniques to identify skin rashes, and dental plaques to X ray analysis and other sorts of visual inputs. 

This saves the clinician’s time in addressing queries and creates a resource that can provide genuine medical advice in times of need. It solves the problems of access, affordability and awareness. Media campaigns to promote the use of such bots can be done to let more and more people become aware of such solutions. 

Such technologies can also be used in routine days by consumers to fill in their history while they are waiting for their turn in waiting areas of regular OPDs. 

A 3-step solution process can be where bots become the 1st point of contact to triage and guide patients, organise them towards the right flow for proper care, followed by Telemedicine to prescribe the patients and then a hospital visit if needed. This practice has been recently adopted by prestigious Mayo Clinic in the United States which can be easily replicated here as well (6). 

These technologies have an added advantage of promoting storage and management of healthcare ‘DATA’ in a digital form. We are sitting on an ocean of data that we are deficient in utilising for nation building right now. So much of our work is not digitised yet. Technologies like these can turn India into a ‘Data’ Gold Mine. As Mr. Kai-Fu Lee of Sinovation Ventures CEO, said ‘In the Age of AI, data is the new oil and China is the New Saudi Arabia’(7). We must make sure we don’t miss the bus by letting go of this priceless resource in the drain. We should be exporting these solutions and not importing those which China/USA has created, tested and approved. We should actively take care of our data and process or utilise it the best we can. In addition to these, there are certain more benefits and many potential applications that are beyond the scope of this article.

Having said all this, there are some limitations that this technology has. All AI models as of today are trained to identify or make a decision out of a fixed set of outputs. Outputs that are limited and pre-determined. An AI Model, trained to identify cats and dogs will not be able to identify a Zebra and will label it as either a cat or a dog, albeit with a low confidence score. Similarly, if a chatbot is trained to identify and guide on say, 100 diseases, it won’t be able to do well on some 101th disease. Or if an X-ray model is designed to identify 10 different kinds of findings in a chest X ray, and there is a 15th finding, it may not give correct results. All these limitations can be solved with more data, expanding the size of outputs and some other research techniques. But these limitations have been dwindled by the zini chatbots as it is able to identify and work on 300+ conditions and is a better option if compared to a local chemist or untrained healthcare provider. And even if the differential diagnosis is not correct, the doctor can still fix the issue as he has the final say in prescribing treatment options. 

I would like to conclude by calling upon the concerned Government bodies to play a proactive role in understanding and implementing these new technologies by drafting policies that promote research and innovation in the field which can quickly transfer tested technologies and solutions for use of the common man. Future is this.

Author:

Dr. Rohit Sharma

MBBS, eMDP IIM Rohtak

Founder GRAINPAD

Co-Inventor of ZINI.ai

References

1. Mayanti M. India Has Only 1 Govt Doctor Per 11,000 People. News 18 [Internet]. 2019; Available from: https://www.news18.com/news/india/only-1-govt-doctor-per-11000-people-is-ayushman-bharat-the-answer-1858705.html

2. Chandra S, Patwardhan K. Allopathic, AYUSH and informal medical practitioners in rural India – a prescription for change. J Ayurveda Integr Med [Internet]. 2018;9(2):143–50. Available from: https://doi.org/10.1016/j.jaim.2018.05.001

3. Sharma M. National Medical Commission Bill 2019 may create 3. India Today [Internet]. 2019; Available from: https://www.indiatoday.in/india/story/national-medical-commission-bill-2019-may-create-3-5-lakh-registered-legal-quacks-in-india-says-doctors-1573249-2019-07-25

4. Chattabotz. What Is A Chatbot? | Chattabotz [Internet]. Available from: https://www.chattabotz.co.za/what-is-a-chatbot

5. André M, Brice P, Cazals D, Hennequin C, Fermé C, Kerneis Y, et al. Results of three courses of adriamycin, bleomycin, vindesine, and dacarbazine with  subtotal nodal irradiation in 189 patients with nodal Hodgkin’s disease (stage I, II and IIIA). Hematol Cell Ther. 1997 Apr;39(2):59–65. 

6. Mayo Clinic Assists At-Home Patients With Voice-Powered Chat [Internet]. Available from: https://patientexperience.wbresearch.com/blog/mayo-clinic-google-assistant-voice-powered-web-chat-strategy-health-wellness-information-to-at-home-patients

7. CNBC Television. Data is the new oil, and China is the new Saudi Arabia, says AI expert [Internet]. YouTube; Available from: https://www.youtube.com/watch?v=EGqg8n3jG7g&t=46s&ab_channel=CNBCTelevision