A chatbot is ‘a computer program that can hold a conversation with a person, usually over the internet’ [1]. It uses artificial intelligence (AI): the use of computer systems that can perform tasks that typically require human intelligence [2]. The development of machines simulating humans is creating a paradigm shift to effectively solve problems in a variety of sectors, including marketing, sales, finance, and healthcare [2,3].
AI chatbots in healthcare can be extremely useful, as they can be accessed 24/7, making it easy for patients and HCPs to get rapid responses to their queries [4,5]. Initially a niche service, healthcare chatbots are now being used much more widely, and in 2020, WhatsApp worked with the World Health Organisation (WHO) to make a chatbot service to answer users’ questions on COVID-19 [3]. Chatbots currently have many applications in healthcare, and it is predicted that the range of uses will continue to increase this decade. AI spending in the healthcare and pharmaceutical industries in 2019 was $463 million, and this figure is expected to grow to more than $2 billion over the next 5 years [4].
Although it’s often thought that chatbots are a recent innovation, the first chatbot used in healthcare is now over 50 years old. It was named ELIZA and was developed in 1966 to imitate a psychotherapist by using pattern matching and response selection [3,6]. However, ELIZA had limited use due to its insufficient knowledge and lack of communication abilities. Chatbots are now more capable and have wider, more flexible applications, such as diagnosing symptoms, and providing mental healthcare consultations and nutrition facts [3]. Chatbots take some pressure off HCPs as they can be used to answer relatively basic enquiries, thus allowing medical information staff more time to handle more complex questions that cannot be readily answered by chatbots. However, the integration of AI into chatbots means that the more questions a chatbot is asked, the more information it can analyse. Also, the quality of the chatbot’s answers can be improved over time, and it can be trained to answer increasingly complex questions [5]. This data also provides insight into the needs of HCPs and patients so when implemented properly, chatbots can help care providers to surpass patient expectations and improve patient outcomes [4,5].
Additional major advantages of using chatbots to interact with users on online platforms is that they can provide anonymity, real time interaction, the ability to interact with numerous users, and they allow monitoring of more patients than a single medical staff member could [3]. In particular, the anonymity of chatbots is preferable for many people when they need to talk about sensitive subjects. For example, the chatbot ‘Woebot’ uses cognitive behaviour therapy and allows anonymous messaging to help people manage their mental health [7]. AI is helping patients self-manage their conditions, and it is also a useful add-on support for pharmaceutical companies. Although anonymity is an important quality of chatbots, the lack of human interaction and delivery can pose a problem to some people who respond best when they interact with another person rather than an AI-driven bot, and there is concern that AI cannot understand the emotional state of a patient, limiting its usefulness [8]. A recent study found that chatbots in healthcare are considered most helpful when their responses match those of a human [4].
Chatbots have been shown to be successful and popular among users, and a recent survey indicated that over 85% of customers would rather get answers from a chatbot than complete a website form [3]. In addition, over 80% of 16-to-24-year-olds would like to see more use of chatbots in the NHS, showing that there is demand for instant healthcare support from the UK young adult population. [9]. The efficacy of using chatbots to screen for heritable cancer syndromes in patients undergoing routine colonoscopy was recently investigated. It was found that the collection of family histories by a chatbot could function as genetic counsellor extenders in busy clinical areas, which enabled the identification of patients with a hereditary cancer syndrome [10]. ‘Ava’, developed by Norgine, is the first patient chatbot by a UK pharma brand. It can answer patients’ questions about their colonoscopy procedure and help them prepare, ultimately resulting in more successful colonoscopies [11]. Therefore, using chatbots can improve cancer diagnosis and ultimately save lives.
Chatbots clearly have many uses in healthcare, and it is likely that further applications will be developed in the future. This will be driven in part by the COVID-19 pandemic, which has increased awareness of the importance of digital technology and steered many away from traditional face to face meetings. Chatbots pose an opportunity of reducing pressure on HCPs by carrying out repetitive administrative tasks, and they work to benefit the public by providing anonymity and instant responses. They do have limitations and cannot completely replicate human interactions, but when used alongside traditional HCP consultations, chatbots can offer HCPs more time to provide higher quality, personalised, and empathetic care to their patients.
Ella White
Bibliography:
[1] Oxfordlearnersdictionaries.com. n.d. chatbot noun – Definition, pictures, pronunciation and usage notes | Oxford Advanced Learner’s Dictionary at OxfordLearnersDictionaries.com. [online] Available at: <https://www.oxfordlearnersdictionaries.com/definition/english/chatbot> [Accessed 6 July 2021].
[2] Builtin.com. n.d. What is Artificial Intelligence? How Does AI Work? | Built In. [online] Available at: <https://builtin.com/artificial-intelligence> [Accessed 6 July 2021].
[3] AIMultiple. 2021. 6 Chatbot Applications / Use Cases in Healthcare in 2021. [online] Available at: <https://research.aimultiple.com/chatbot-healthcare/> [Accessed 6 July 2021].
[4] Technology Solutions That Drive Healthcare. 2020. “Are You There, Chatbot?”: Automated Care Grows Up. [online] Available at: <https://healthtechmagazine.net/article/2020/08/are-you-there-chatbot-automated-care-grows> [Accessed 6 July 2021].
[5] Babylon Health. n.d. AI guide in healthcare. [online] Available at: <https://www.babylonhealth.com/ai/guide-in-healthcare> [Accessed 6 July 2021].
[6] Analytics India Magazine. 2016. Story of ELIZA, the first chatbot developed in 1966. [online] Available at: <https://analyticsindiamag.com/story-eliza-first-chatbot-developed-1966/> [Accessed 6 July 2021].
[7] PMLive. 2019. AI is transforming pharma R&D – now it can transform pharma marketing. [online] Available at: <http://www.pmlive.com/pharma_thought_leadership/ai_is_transforming_pharma_r_and_d_now_it_can_transform_pharma_marketing_1297189> [Accessed 6 July 2021].
[8] Palanica, A., Flaschner, P., Thommandram, A., Li, M. and Fossat, Y., 2019. Physicians’ Perceptions of Chatbots in Health Care: Cross-Sectional Web-Based Survey. Journal of Medical Internet Research, [online] 21(4), p.e12887. Available at: <https://www.jmir.org/2019/4/e12887/PDF> [Accessed 6 July 2021].
[9] Earthware.co.uk. n.d. [online] Available at: <https://earthware.co.uk/ai-white-paper.pdf?v=1> [Accessed 6 July 2021].
[10] Heald, B., Keel, E., Marquard, J., Burke, C., Kalady, M., Church, J., Liska, D., Mankaney, G., Hurley, K. and Eng, C., 2020. Using chatbots to screen for heritable cancer syndromes in patients undergoing routine colonoscopy. Journal of Medical Genetics, [online] pp.jmedgenet-2020-107294. Available at: <https://jmg.bmj.com/content/early/2020/11/08/jmedgenet-2020-107294> [Accessed 6 July 2021].
[11] earthware. n.d. [online] Available at: <https://earthware.co.uk/case-studies/colonoscopy-support> [Accessed 6 July 2021].
