Artificial intelligence (AI) is transforming the way healthcare organizations can deliver remote patient monitoring at scale and improve patient outcomes.
Look no further than the COVID-19 crisis to understand the value that conversational AI solutions and virtual assistants can bring to healthcare institutions. As the number of infected cases rise exponentially, hospitals get overwhelmed with patients, at varying levels of severity and risk. Northern Italy saw the coronavirus strain their healthcare system in three weeks, with doctors forced to decide whether or not to treat some of the elderly patients. Earlier in China, patients without the virus were turned away by hospitals, including those needing surgery for tumours. Similar stories were reported in Jakarta where hospitals were scrambling for tests and in Manila where doctors had to stop accepting patients due to surging COVID-19 cases.
The Root of the Problem: Prioritisation
The speed at which the infections multiply is definitely a concern. But the deeper problem at the heart of the crisis has to do with the lack of resources, ventilators and other medical supplies to meet the exponential growth in patients. The resulting physical and mental stress placed on healthcare staff only makes matters worse. If the number of patients remained steady for say, a month, providing necessary care to critical patients until discharge would be relatively more manageable. Our healthcare institutions have not been designed to scale for such an exponential growth in patients.
“Our healthcare institutions have not been designed to scale for such an exponential growth in patients.”
So as the number of infected patients rise, they rapidly use up the hospital’s resources – both in the form of available healthcare workers and medical equipment. For example, some may merely want to get a pre-emptive test for reassurance while others with mild symptoms rightfully will need quarantined care. Those in critical conditions on the other hand will require Intensive Care with ventilators and emergency life-prolonging devices.
This is in addition to other patients who may be in for general treatment and surgeries for less serious diseases or even routine care. In some cases, the volume of infected patients going into hospitals has been so high that healthcare staff and doctors themselves were exposed, exacerbating the crisis even more.
Conversational AI for Triaging Patient Treatment
This is where a conversational AI agent can make a big difference by streamlining the screening process. By automatically separating patients into different risk levels, the high-risk cases can be expedited for priority attention without the lower risk patients taking up valuable resources in the system. Deployed on the website of the hospital or clinic, or their Facebook or WhatsApp business account, these virtual assistants automate a first level of support and triaging. This decreases the workload on customer service teams, reception staff and nurses while customers can converse directly with these bots regarding:
- Scheduling, rescheduling and cancelling appointments
- Frequently Asked Questions like opening hours, walk-in appointments, clinic locations
- Type of care based on age, gender and other attributes as well as type of symptoms
- Patient registration
Consider three scenarios specific to the COVID-19 crisis. Three different people are seeking information on the right next steps for them.
- Individual A: 32-year old, anxious after arriving from a work-related trip to an affected country, wants to check if he is infected.
- Individual B: A 69-year old with a pre-existing condition of asthma has been feeling feverish, out of breath and coughing regularly for the last two weeks.
- Individual C: A 15-year old living in a COVID-19 affected cluster but with no symptoms wants to get tested.
Any other time, it would be reasonable for all three to schedule an appointment at a clinic to consult a doctor and even take the COVID-19 test. But considering the severity of the crisis and the huge number of people who may roughly fall into the above 3 categories and the danger of transmission in large crowds, it might be wise to provide the following responses:
- Individual A: Self-quarantine at home for 2-3 days while keeping distance from others, especially the elderly and practicing good hygiene including frequent washing of hands with soap. If symptoms persist after 3 days, make an appointment with a General Practitioner wearing a surgical mask.
- Individual B: Report immediately to the emergency department at the nearest. hospital, wearing a surgical mask and preferably in a cab (avoiding public transport)
- Individual C: May not have COVID-19 but would be wise to self-monitor for any symptoms.
“Segmenting patients automatically based on the severity of the case allows healthcare staff to prioritise and focus on the high-risk cases more effectively”
Segmenting patients based on the severity of the case like above allows healthcare staff to prioritise and focus on the high-risk cases more effectively. And with all of these instructions automated by a conversational AI agent, reception and counter staff only have to deal with those who fit into the Individual B category at a given time.
Learnings from the Field
We at KeyReply are already seeing the impact such conversational AI solutions can have in healthcare. In various implementations with a major healthcare provider and public healthcare institutions in Singapore, our chatbot, deployed on the hospitals’ webpages and social media channel, helped customers with answering queries regarding the symptoms they observed and what to do about it, appointment booking and provision and dissemination of information on health and wellness content. We observed that:
- More than 60% of users expressed intention to make appointments or wanted to visit a doctor
- As the number of interactions with the conversation AI agents doubled, the requests for appointments increased at a higher rate
- The descriptions and terms used by patients and their caregivers tend to be generic and are quite different from the professional terminologies and definitions used by clinicians.
This means that there is tremendous potential in managing the appointment request use case from a business and operations management perspective in private and public healthcare institutions respectively.
Also, due to the high impact nature of the healthcare sector, it is more important to be correct, or in an uncertain situation, to be more conservative. The emphasis should also be to match the patients’ descriptions of the symptoms to how the clinicians would interpret them. Mapping out detailed clinical protocols is important in ensuring accuracy of the triaging logic.
Our studies also revealed that the top 5 intents for questions asked of the chatbots included doctor and specialist recommendations, symptoms checking, appointment related questions, billing and those regarding general health and check-up services.
Such enquiries and tasks pertaining to operations and logistics are ripe for automatic resolution through conversational AI agents. With these matters off their minds, healthcare staff can focus on more value adding and lifesaving tasks, caring for priority patients and strengthening customer support.
Towards the Future of Healthcare Institutions
The same patient-facing applications of conversational AI can be applied across other functions within the healthcare institution to smoothen access to information and data for all teams. From back-end process optimisation to pharmaceutical research and even medical diagnosis, AI technologies are already showing us glimpses into the healthcare of tomorrow. Soon, care teams will be able to quickly access daily patient caseload, check for cancelled appointments and request specific reports – all by conversing with an intelligent bot. Meanwhile the same bot could help management teams pull up info on patient satisfaction rating, claims data and monthly admission rate.
Conversational AI is also becoming integral to discussions around how outpatient care can be administered at scale with technology. This involves novel approaches by healthcare institutions to improve the quality and flow of their outpatient services, reducing friction pre-appointment and waiting times for patients as well as caring for them with post-treatment follow-ups. Today’s healthcare institutions also invest significantly more resources into treatment than preventative care. A lot of this can be saved if patients are contacted on a recurring basis to check in on their health so that any major illnesses can be prevented by following the right routines, medications and prescriptions.
KeyReply’s solutions can help healthcare institutions around:
- Symptoms matching and triaging – where patients are guided to the right doctors based on an analysis of the symptoms that they enter into the chatbot
- Appointment making and actualisation – an end-to-end management of administrative tasks for the patient, from appointment scheduling, registration, treatment and financial counselling post-treatment as well as back-of-office operations for billing and access to medical reports
- Pre-procedure operations – streamlining health screening by automating queries regarding fasting, medications, allergies and protocols before appointment to help front–of–office operations go smoothly on the day
- Outpatient care – Regular follow ups to check blood pressure, diabetes and cholesterol readings, with notifications corresponding to the health status of the patient.
Consider the numerous additional supporting functions within a healthcare institution where conversational AI solutions can be applied, like retrieval of medical protocols, Human Resources, Finance, IT Help Desk and Procurement. The benefits to internal productivity and organisation-wide growth are immense. And the opportunities endless.
Find out how KeyReply can help streamline and prioritise patient screening for you.