An overview of chatbot technology in healthcare institutions
With exceptional growth in Artificial intelligence, chatbots have begun to rule various industries. One major contribution of a chatbot is in healthcare. Communication and organization became easier with the know-it-all chatbots. The key focus is that these chatbots are available whenever and wherever one requires them. The idea is quite simple, a system generating automated responses based on the user interaction. The responses are stored in a database. It tries to provide the most likely cause of the users' symptoms and can even measure the severity of the disease. If at all the condition requires further assessment, the chatbot immediately connects the user to the next available physician. One key feature of a chatbot is that it maintains a human-to-human relationship with the user. This paper has highlighted the advantages of implementing chatbots in the healthcare sector. Medical chatbots can connect with patients efficiently and provide first-aid services without any delay.
We are living in the 21 st century and we expect anything and everything happening around us to be nothing less than accurate, rapid, and efficient. Especially in a highly populated country, each of us requires equal attention. This becomes challenging in large sectors such as the healthcare sector where immediate care is necessary. The healthcare sector is huge in terms of employment and economy. Chatbots are the latest development that has the potential to revolutionize healthcare.
For over decades, we have come up with various advances in technology. During the 1990s when telephones were a thing, telephone-linked care was given to people. When mobile phones such as Nokia came into the picture, SMS-texting intervened and became quite a success. Now, in a time where the internet and smartphones are ruling, chatbots and other similar applications have gained the spotlight. 
Now, what exactly are chatbots? They come under machine learning which comes under Artificial Intelligence. AI systems can enhance decision-making and have the ability to imitate human cognitive functions. They are designed in a specific way to provide a real-time experience with the end-user. It is simply a computer program that engages the user in the form of text messages within the website or application. This technique is now widely used in analyzing symptoms, managing medications, and even monitors chronic diseases. What needs emphasis is that these systems aid physicians with accurate medical information to reduce error in diagnosis and even alert them about a possible high-risk disease which could help in providing the patient with appropriate and immediate care.
Chatbots can increase the communication between patient-doctor and patient-clinic. Not only does it help in setting up personalized health matters, but it can also monitor behavioral indicators such as nutrition, sleep, and physical activities. Such technologies prove to be cost-effective and can decrease the gap between health and well-being. Something that stands out is that chatbots give instant responses to patient inquiries according to the symptoms presented while looking for specific patterns in predicting diseases. These can be modified based on the population, age group, or already existing health conditions.
Moreover, the anonymous nature of a chatbot makes it a reliable source of information with regards to the speed and accuracy of the information received. In a study, it was concluded that the millennials preferred texts over calls. This means that these chatbots already have an edge in the modern world.
Flora Amato  discussed stimulating human interactions in the medical context with the help of chatbots. It is capable of overcoming classical human-machine interactions. Thus allowing the patients to communicate freely. The paper specifically talks about HMOLeS (Health On-Line Medical Suggestions), a system that can adapt to different medical scenarios. It was designed to be modular, where each concern is associated with a given module. This makes it fulfill all the desired characteristics of a medical chatbot.
Benilda Eleonor  The paper proposed a system-A Pharmabot-that can be used by parents of children to determine the medical assistance in taking the generic medicine for children. They used pictorial representation to put forth the idea. Also the Left and Right Parsing algorithm to obtain the desired outcome.
Divya S  In this paper, they discussed using text-to-text communication with the user using the chatbot. This enables the user to get specific details about the disease and also their previous chat history which is stored in the database. It collects the symptoms and does symptom mapping to classify them as major or minor diseases.
T Kowatsch  focused on an existing instance of a text-based chatbot that was designed to support both the patients and healthcare professionals beyond on-site consultation. It demonstrated how the patients can chat with the doctor daily and the doctor can monitor the patient more carefully. The overall satisfaction of the system was proven to be good.
Mohammed Javed  is his paper proposed an algorithm for word segmentation. He devised an algorithm to calculate character spaces in sentences. A sentence included all the punctuations and gaps. After detecting the character spaces, the mean average is calculated. Based on the position of characters, their compressed segments are extracted. The gap between two words is always greater than the average character space.
LinHua Gao et al.  The paper talks about the method of synonym extraction. The system must understand the user requirement and act accordingly. The system scans for the keyword and it is necessary to have a dataset of synonyms of that keyword. All possible synonyms are matched and the process continues.
Advantages of implementing chatbots
Most of us know how beneficial technologies can be. And something as advanced as a chatbot can only make things easier. A quick countdown of its advantages would make us want to choose chatbots over every other method.
1. 24/7 availability
Unlike doctors and other medical professionals, chatbots work 24/7 without complaining or tiredness. The database is stored with all information related to health and is provided as in when requested. These AI-controlled chatbots assist you throughout your recovery. Be it as a reminder for taking your medications or keeping overall monitoring of your health.
2. Instant information and medical history
Time plays a crucial role in healthcare. Every second is valuable. During an emergency, the time you delay may cause something unfavorable. Healthcare chatbots can save time because the doctors are provided with medical histories, allergies, and every other medical detail about that patient in an instant.
3. Guides the user
One of the most important qualities any program needs to attain is user-friendliness. Things in front of their screen should be clear and understandable. Chatbots appear as pop-up screens which can immediately catch attention and provide further guidance.
4. Schedule appointments
As discussed earlier, booking appointments become hassle-free with chatbots. From checking the doctor's availability to marking the slot in both the patient's and doctor's calendars, chatbots do them efficiently.
5. Check for symptoms exactly how a physician does
Chatbots behave exactly how a doctor behaves during the first examination. That is, asking for symptoms. Chatbots have a well-structured question bank (usually asked by doctors) before checking the severity of the symptom and coming to a conclusion. Also, entering your details each time can be ruled out.
6. A source of information
There are always people who call for additional inquiries such as insurance claims or as simple as repeating what was said. In a busy atmosphere, these doubts kill time. A chatbot manages it all. It Saves time provides information and reduces the number of repetitive calls.
7. Reduces cost
Hospitals tend to charge for every test, diagnosis, doctor visits, follow-ups, medications, and so on. Chatbots can cut down a huge amount of the initial assessments.
8. Patient feedback
Collecting feedback is an age-old practice. Most people ignore such emails or other forms through which they are required to fill feedback until and unless there's a drawback. Chatbots in this scenario can be quite useful. People find these a convenient way rather than taking extra time to fill big forms.
9. Automated invoicing and payments
The medical billing process is laborious and time-consuming. Chatbots can be programmed to send an invoice and collect payment. This can also reduce the administrative charges.
10. Improve internal communication
Not only with the patients, but chatbots can also be used between employees to facilitate better communication. Any information can be obtained from their site.
The function of every chatbot is similar. It extracts personal information (like name, phone number) and the symptoms causing trouble. They store this information in a database to facilitate patient admission, symptom tracking, and medical records.
Firstly after initiating the conversation, the chatbot gathers the keywords from the first interaction. After getting a brief scenario, the chatbot takes control of the conversation by asking questions more specifically. This leads to a possible number of diseases which are then ordered. After getting a clear picture, the chatbot will be able to finalize a disease. It then checks the severity of the disease and either prescribe medications or connects the patient to a doctor who could give better care. The following will demonstrate a sample chat of two different chatbots,
Chatbot1: Hi! Enter your symptoms in not more than 300 words.
User: I have throat pain.
Chatbot1: Check the following questionnaire asked by the doctors in a similar situation. <list of questions>
Chatbot1: If your answer isn't in the list above, please send an SMS “send” to 800145.
Chatbot2: Hey! How are you feeling?
User: I have a throat-pain
Chatbot2: Do you also have a fever?
User: Yes I do.
Chatbot2: Have you been in contact with a covid patient in the last 14 days?
User: Yes, I have.
Chatbot2: Please come in for a covid-19 test and isolate yourself to be on a safer side. I have booked the appointment for tomorrow at 9 am.
Chatbot2: You could take a Panadol to bring down the fever and also a Prospan-syrup for the throat pain.
Chatbot2: Is there anything else I can help you with?
User: No, thank you!
The difference between the two chatbots: one does not provide the cure. It only provides the questions previously asked by a physician to find something common. The other one gives the medication for the exact symptom.
The main aim of a medical chatbot is to assist 24/7. They try to create a chatbot that can empathize with the patient and then proceed to give medical information by communicating in the natural language. The chatbot initiates the conversation and moderates it. Some of the advanced forms can detect voice messages, facial expressions, and other movements.
Some of the medical chatbots used web-based text messaging applications where it delivered the sequence of how the conversation between a patient went. Another chatbot system gathered input and processed the data containing medical terminologies and provided the output solutions based on it.
To understand how a typical chatbot operates, one must explore its implementation strategies. An overview of how a chatbot works can be explained. A chatbot is programmed to behave like a human in our natural language. As soon as the user inputs a message, the chatbot gets activated and retrieves the messages sent by the user. The system scans the data and extracts the keywords. The keywords are then cross-checked with the disease tags already in the database and come to more than one conclusion. Next, the system narrows it down to the specific disease with the help of questions or if in case more than one disease is likely, the user is connected to a doctor. Once the possible disease is confirmed, it measures the seriousness of the disease. The system maintains an integer variable where it sums up the scores of the symptoms if it matches with the user input.  The chatbot also has other arrays ‘medications’ and ‘remedies’. When the symptoms match, the corresponding medications are stored in the side of the array by side. After the symptom checks are done, the information stored in the arrays is provided as an output to the user.
Some of the fundamental concepts implemented in a chatbot are as follows:
a) Pattern Matching: An user enters a sentence and the output is generated in accordant with the user input. ELISA was one of the first chatbots to use this technique i.e, pattern recognition. The drawback of this method is that it lacks a human touch. The results are either repetitive or predictable. There is no storing of data which may lead to beating around the bush.
b) AIML: Artificial Intelligence Markup Language is based on pattern recognition. It follows the user input-response approach where natural language is used to facilitate the exchange of dialogues between a human and the chatbot.
Pattern Detection in AIML
AIML is an XML-based markup language and is tag-based. AIML is based on basic units called categories which are formed by the pattern inputted by the user and the responses by the chatbot.
<pattern> My name is * </pattern>
<template> Hello, please tell me know your symptoms. </template>
Pattern Detection using Snippets
The chatbot focuses on the keyword provided by the user. This then is matched with the pre-defined pattern, and if so it matches, further steps are taken. There can be two scenarios of how a chatbot detects.
The user knows their disease and just wants to know its medications.
Which medication should I take for [*]
Here, the [*] represents the disease the patient is suffering from. The chatbot provides the data in [*] to the system and the remedies to cure that particular disease will be provided to the patient.
The user does not know the disease and provides the initial symptoms to the chatbot.
I have a bad [*]
Here, the [*] represents the symptom like a cough. The chatbot extracts the data in the [*] to the system and examines the patient further to conclude.
Similar to pattern detection, two main types of categories are established in a chatbot.
When the user wants only the remedies.
When the user wants to find out the disease.
The Chatbot engines interact with the backend file which contains the medical data provided by the medical professionals. The input can either be a symptom the patient is suffering or the disease name.
Medical Data Storage, Shortlisting and detecting a disease
Storing data is usually a tedious task. Different symptoms, the types of diseases, and their remedies have to be stored in an orderly manner. The chatbot uses an XML sheet to store them.
When the user interacts with the chatbot, the system removes the unnecessary punctuations and assigns scores to disease. The lower scores are disqualified as the tags of the disease do not match with the symptoms provided.
After shortlisting several diseases, it has to find out the exact one. This is done by further examining the user by asking various questions. When they match, scores are provided again. If in case the chatbot is unable to fix a disease, the patient is connected directly to a physician.
Once the chatbot identifies the disease, it has to provide medications for the same. The chatbot further uses the medication and remedy array to provide the final solution.
Result and Discussion
A chatbot has the ability to revolutionize healthcare with its minimal human interaction. A chatbot works on the principle of artificial intelligence. The communication between the user and the system is text-based. The data is stored in the database and it is retrieved as and when required.
Since these chatbots are available 24*7, medications can be provided in case of an emergency or an unfavorable situation. They are a reliable source of medical information.
As technology continues to rule the world, it is time that people shift to medical chatbots. Medical chatbots have the advantage of being around whenever required. This means one does not have to compromise their busy schedules to treat some minor illness. They are a personalized system of medical diagnosis. Chatbots have proven to be user-friendly and anyone with basic skills in language and technology can avail of its use.
AI is the main brain behind a chatbot and it also relies on medical data that have been supplemented. Implementing a chatbot can save lives and alert people in case of an emergency. It is one of the best alternatives in a world where messages dominate person-to-person interaction.
Author: Arun Poochelvan
- Adamopoulou, Eleni, and Lefteris Moussiades. “An Overview of Chatbot Technology.” Artificial Intelligence Applications and Innovations: 16th IFIP WG 12.5 International Conference, AIAI 2020, Neos Marmaras, Greece, June 5–7, 2020. (https://te.1lib.in/book/11252181/e7093b)
- Flora Amato, Stefano Marrone, “Chatbots meet eHealth: automat zing healthcare”, proceeding of diet, May 2018. (https://issuu.com/irjet/docs/irjet-v7i2422)
- Benilda Eleonor V. Comendador, “Pharmabot: a pediatric generic medicine consultant chatbot”, proceeding of the JACE, April 2015. (https://iarjset.com/wp-content/uploads/2021/05/IARJSET.2021.8524.pdf)
- Divya, Indumathi, Ishwarya, Priyasankari, “A Self-diagnosis medical chatbot using artificial intelligence”, proceeding MAT journal, October 2017. (https://www.ijitee.org/wp-content/uploads/Souvenir_Volume-9_Issue-6_April_2020.pdf)
- Tobias Kowatsch, “Text-based healthcare chatbots supporting patient and health”, October 2017. (https://www.researchgate.net/figure/A-text-based-healthcare-chatbot-using-health-data-from-everyday-objects-and-self-reports_fig1_320161507)
- Mohammed Javed, P. Nagabhushan, B.B. Chaudhari, “A Direct Approach for Word and Character Segmentation in Run-Length Compressed Documents with an Application to Word Spotting”, 13th International Conference on Document Analysis and Recognition (ICDAR), 2015. (http://www.ijstr.org/paper-references.php?ref=IJSTR-0719-20726)
- Informatica 31 (2007) 249-268 249 Supervised Machine Learning: A Review of Classification Techniques S. B. Kotsiantis Department of Computer Science and Technology University of Peloponnese, Greece End of Karaiskaki, 22100, Tripolis GR.(https://www.academia.edu/6428653/Supervised_Machine_Learning_A_Review_of_Classification_Techniques)
- LinHua gao, HePing Chen, “An Automatic Extraction Method Based on Synonym Dictionary for Web Reptile Question and Answer” 2018. (https://www.researchgate.net/publication/326050646_An_automatic_extraction_method_based_on_synonym_dictionary_for_web_reptile_question_and_answer)
- Nadarzynski, Tom et al. “Acceptability of artificial intelligence (AI)-led chatbot services in healthcare: A mixed-methods study.” Digital health vol. 5, 21 Aug. 2019. (https://www.researchgate.net/publication/335317322_Acceptability_of_artificial_intelligence_AI-led_chatbot_services_in_healthcare_A_mixed-methods_study)
- Engati Team, “Why healthcare chatbots are trending in 2021”, May 22. (https://www.engati.com/blog/chatbots-for-healthcare)
- Abd-Alrazaq, Alaa et al. “Technical Metrics Used to Evaluate Health Care Chatbots: Scoping Review.” Journal of medical Internet research vol. 22, 5 Jun. 2020. (https://arbor.bfh.ch/view/year/2020.type.html)
- Hephzibah Thomas, “Critical Literature Review on Chatbots in Education”, International Journal of trend in scientific research and development, vol 4, issue 6, October 2020. (https://issuu.com/ijtsrd.com/docs/140_20critical_20literature_20review_20on_20chatbo)
- Nivedita Bhirud, Subhash Tataale, Sayali Randive, Shubham Nahar, “A Literature Review On Chatbots In Healthcare Domain”, International journal of scientific & technology research, vol 8, issue 7, July 2019. (http://www.ijstr.org/final-print/july2019/A-Literature-Review-On-Chatbots-In-Healthcare-Domain.pdf)
- Andrew Reyner, Wibowo, Audrey Chen, Hubert, Edy Irwansyah and Bayu Kanigoro, “Medical Chatbot Techniques: A Review”, Research gate, November 2020. (https://dokumen.pub/software-engineering-perspectives-in-intelligent-systems-proceedings-of-4th-computational-methods-in-systems-and-software-2020-vol1-1st-ed-9783030633219-9783030633226.html)
- Rarhi, K., Bhattacharya, A., Mishra, A., & Mandal, K, “Automated Medical Chatbot”, SSRN Electronic Journal, 2017. (https://www.ijsdr.org/papers/IJSDR2009083.pdf)
- Ashwini, Ankita, Sakshi, Nidhi, Aman, Pankaj, “Artificial Intelligence-based Healthcare Chatbot System”, International Research Journal of Engineering and Technology, vol 7, issue 2, February 2020. (https://irjet.org/volume-7-issue-02)