Improve the therapeutic adherence of cancer patients with a chatbot

Vik

Until a few years ago, cancer treatment was mainly carried out within the hospital. Advances in treatment are making it increasingly possible to treat these diseases as chronic diseases. The home management of these patients presents new challenges that require new tools.

Home medication creates new challenges for patients with cancer, greatly improving their quality of life, but they are responsible for the correct intake of drugs, a situation that is not without risks. Prescription adherence can prove more difficult without a team of caregivers surrounding the patient. Continuing for months without seeing the oncologist may lead to feelings of isolation and many questions may be left unanswered.

This new scenario has led French physicians to work on a solution to better assist patients throughout treatment, by offering a supplement to the conventional doctor-patient relationship.

More and more cancers are being treated like chronic diseases, with higher survival rates and home prescription. Up until a few years ago, cancer care was being entirely provided inside the hospital. We decided to develop a 24/7 available virtual assistant to both answer queries and tackle fears patients may have once they are alone at home” says Benjamin Chaix, scientific director of Wefight.

Wefight is a medical startup which started working on AI-fuelled chatbot in 2017. This conversational robot can understand questions that are sent in text, and deliver answers based on scientifically validated content. The chatbot, called Vik, is available for patients with breast cancer, asthma and depression, and can be used on Messenger.

Vik works with natural language processing (NLP) algorithms and stores all the information on secure servers that have been certified for healthcare data storage.

Vik2

The architecture of the platform is composed of several technology bricks that allow analysing patients queries and sending back personalised answers. The system deciphers the sentence and identifies words, or groups of words, using machine learning techniques, and activates the different modules depending on the words it detects, to prepare adequate content, according to the topic of discussion. Answers in every activated module are then regrouped to issue a personalised answer, which is sent back to users. The platform is trained with clinically validated patient questions that are controlled by healthcare professionals.

Wefight recently carried out the first prospective trial on how patients interact with a chatbot, together with physicians from Montpellier University Hospital Gui de Chauliac and European Hospital Georges Pompidou in Paris. The results are particularly encouraging in terms of treatment effectiveness.
The trial meant that 4,737 patients with breast cancer were enrolled and observed over one year to evaluate their functional and emotional interactions with the chatbot, and quantify their adherence using the assistance in disease follow-up.

On average 132,970 messages were exchanged per month between patients and Vik. Researchers then calculated the average medication adherence rate over four weeks, by using the prescription reminder function, which patients can activate themselves.

Treatment adherence improved by over 20% when using the function. “The more patients used the chatbot, the more adherent they were. Patients regularly left positive comments and recommended Vik to their friends. The overall satistfactions was 93.95%. When asked what Vik meant to them and brought them, 88% said that Vik provided them with support and helped them track their treatment effectively,” Chaix said.
Patients included in the study had breast cancer or were in remission or acted as caregivers to the patients. The average patient age was 48 for 89% women and 11% men. Wefight said it would complete these results with a more advanced and specific study.

Vik brings scientifically validated information under one roof, to help patients find all the evidence they need to know. The chatbot can provide an immediate answer to the patient on practical themes, without having to book on appointment with the oncologist.

The chatbot enables us to refresh answers the physician already gave to the patient during consultation. A lot of information is given during consultation, and it may be difficult for a patient to remember it all” he said.

Typical questions address patient side effects on life quality such as the effect on sports practice, psychology, nutrition, hair loss, sexuality, and if there is any pain involved (e.g. during or after biopsy). Practical aspects, such as how to obtain reimbursement for prosthesis, are also common.
Patient group Patients en Réseau was involved in the development of the chatbot from the start, notably through focus groups, to help define relevant content.

Laure Guéroult-Accolas, founder of the group, highlighted the value of the tool in view of the new possibilities it grants patients. “Discussion with other patients is essential to compare experience and give advice and/or supports. But there might be things you don’t want to share, or you may be looking for very factual information. The chatbot comes without ties, so you can use it as an additional information source” she said.

Ambulatory care offers more autonomy, but also more loneliness. “You get your medication from the pharmacy and are sort of left alone to manage that aspect of your disease. That’s not easy to live with on a daily basis, especially if your treatment lasts 5,7 or 10 years, as for example in hormone therapy. Typically, this kind of treatment also has side effects, which are not so relevant to the physicians anymore, but can heavily impact on the patient’s life – early menopause, flushing, joint paint” Laure Guéroult-Accolas explained.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s