One of the biggest challenges facing the healthcare industry is due to constraints such as limited doctor office hours and limited education about places of care (such as This article). The Covid-19 pandemic has further exacerbated accessibility challenges as care resources become more scarce. Government agencies and healthcare organizations have begun to use or explore how technology can help improve patient access and ease the burden on care providers. One such technology is chatbots.
While medical chatbots have a wide range of applications, such as providing medical information and mental health assistance, typical chatbot solutions are often inadequate, frustrating both users and owners. Expected care accessibility is lost due to chatbots ignoring or misinterpreting user queries or failing to complete assigned tasks (e.g., care triage). It also undercuts the original purpose of reducing the burden on care providers if it takes a lot of time and resources to set up and maintain a chatbot. There may be hundreds of chatbot platforms on the market, and here is the challenge – How to choose a chatbot platform for healthcare?
Considering improving patient access to care, government agencies and healthcare organizations should look for chatbot platforms with three characteristics: quality of conversational engagement, ease and speed of chatbot setup, and continuous, non-disruptive chatbot improvement.
- Quality of Dialogue Participation
Unlike many other service industries, healthcare provides services that make people feel better physically and mentally. When a chatbot interacts with visitors and patients on behalf of a care facility, it must be instilled with a sense of responsibility and empathy, just as visitors and patients interact with real people. In other words, chatbots must not only have conversations, they should also have high-quality and fruitful conversations with users, increasing access to care. Below are a few examples that demonstrate the characteristics of quality and productive interactions. Chatbots that can support such high-quality conversations are also commonly referred to as cognitive AI chatbots.
Actively listen to users
In order to have productive conversations with users, an effective chatbot should actively listen to users and respond to user requests responsibly and empathetically, rather than forcing users to follow predetermined, rigid paths to navigate care options. The following example shows a user interrupting an urgent care triage chatbot by asking a question. In this case, the chatbot quickly answers the user’s question and then continues the conversation.
|
Interactions between urgent care triage chatbots and users. During the classification process, the user interrupts the process with a question. The chatbot is able to explain the problem and respond quickly before continuing the process. |
The patient was already in a very stressful emotional and physical condition when he/she inquired about nursing services.Like an empathetic care provider can help relieve stressful situations and provide better care outcomes (see This article Regarding the role of empathy in healthcare), healthcare chatbots should also feel how patients are feeling and respond with empathy. The example below shows an interaction between a chatbot and a patient during a mental health assessment. Chatbots are able to actively listen and respond empathetically to users.
|
Chatbot interaction with the patient during the mental health assessment process, during which the chatbot actively listens to the patient and responds to her empathetically. |
read between the lines
Study Shows Patient Personality Correlates with Medication Adherence for Both elderly and teenager, which indicates the need for an individualized care management program. When using chatbots as part of a care management scheme (for example, monitoring patient status and encouraging treatment adherence), such chatbots should also be able to gain insight into each user and use insights to personalize each engagement.
The example below shows a personal wellness chatbot interacting with two different users and encouraging the two users to stay in a wellness program with different motivational statements based on the users’ personalities (inferred by the chatbot during their respective conversations) .
|
Healthcare chatbots are able to infer personality insights from conversations and use those insights to personalize each engagement. On the left, the chatbot interacts with a motivated, independent person, while on the right, the chatbot interacts with a caring, family-oriented person. |
|
To support the deeply personalized interactions shown above, healthcare chatbots must analyze user input beyond the surface to automatically infer the user’s unspoken needs and desires and personality. The chatbot can then use the inferred insights to personalize each engagement.Usually such an inference is given by Combining Big Data Analysis with Modern Psychometric Theory. Field studies have also shown the usefulness of this inference Predict team performance.
- Ease and Speed of Chatbot Setup
If a chatbot platform claims to provide the quality of conversations described above, it is important to ask how easy and fast it is to set up and deploy such a chatbot in a production environment. This is because building a high-quality chatbot from scratch requires AI expertise and a mature IT team, not to mention large amounts of training data and intensive computing resources. Many organizations underestimate the importance of value time associated with chatbot projects and the importance of using time value to evaluate chatbot platforms. Otherwise, assuming they ever get successful results, they may be very disappointed with the actual timeline required or frustrated with the quality of the chatbot because they spent a lot of time building it.
Since few healthcare organizations have the required AI/IT expertise or resources, it is important to choose a chatbot platform that can support end-to-end, no-code design, development, and deployment of chatbots, which also provides high Quality conversational engagement, such as active listening skills.
Also, it is important to check if the no-code chatbot platform is enabled Reusable Artificial Intelligencewhich will greatly speed up the setup process, as chatbots can be quickly customized using pre-built AI modules (for example, a pre-built active listening engine) without having to build everything from scratch.
- Continuous, uninterrupted chatbot improvements
No chatbot is perfect, it requires human oversight and continuous maintenance and improvement. In addition, healthcare situations can change rapidly, such as pandemic situations and corresponding care policies. On the other hand, every healthcare-related conversation is a critical conversation, whether a patient is asking about treatment options or undergoing a health assessment. Healthcare organizations simply cannot interrupt any ongoing conversations to teach their chatbots new knowledge (for example, informing patients of new care policies during a pandemic) to improve their capabilities.
Therefore, it is also important to choose a chatbot platform that can support continuous, non-disruptive improvement.For example, if a user asks a question that a medical chatbot cannot answer, the platform should notify the chatbot’s human supervisor in real-time and let him/her improve the chatbot immediately no Interrupt any ongoing conversation.
Summary: Due Diligence Questions
In general, healthcare services are people-centred services. When healthcare organizations choose chatbot solutions to grow their workforce and improve patient access to healthcare, it is important to choose a chatbot platform that enables human-centric engagement at scale and features 3 Prerequisites: (1) Provide high-quality and efficient engagement, (2) Set up quickly and easily, (3) Support continuous, non-disruptive chatbot improvement.
When considering or evaluating a healthcare chatbot platform, use the following questions to conduct due diligence:
Evaluate conversation quality
- I want to use a chatbot for patient interaction. How does it handle arbitrary user interrupts? How does it resume chat flow from user interruptions?
- How does it handle user free text issues?
- How does it handle complex issues that require multiple rounds of conversations?
- What can it learn about users through conversations?
Evaluate ease and speed of setup
- How fast can it be set up?
- through whom?
- What resources do I need to do this?
Evaluate Chatbot Improvements
- How do I know if my chatbot has made mistakes or can’t answer certain questions?
- How can I improve my chatbot?
- Do I need to close my chatbot or interrupt an ongoing chat to improve it?
Photo: venimo, Getty Images
