The Development of a Chatbot Technology to Disseminate Post–COVID-19 Information: Descriptive Implementation Study
Bibliographic record
Abstract
BACKGROUND: Post-COVID-19, or long COVID, has now affected millions of individuals, resulting in fatigue, neurocognitive symptoms, and an impact on daily life. The uncertainty of knowledge around this condition, including its overall prevalence, pathophysiology, and management, along with the growing numbers of affected individuals, has created an essential need for information and disease management. This has become even more critical in a time of abundant online misinformation and potential misleading of patients and health care professionals. OBJECTIVE: The RAFAEL platform is an ecosystem created to address the information about and management of post-COVID-19, integrating online information, webinars, and chatbot technology to answer a large number of individuals in a time- and resource-limited setting. This paper describes the development and deployment of the RAFAEL platform and chatbot in addressing post-COVID-19 in children and adults. METHODS: The RAFAEL study took place in Geneva, Switzerland. The RAFAEL platform and chatbot were made available online, and all users were considered participants of this study. The development phase started in December 2020 and included developing the concept, the backend, and the frontend, as well as beta testing. The specific strategy behind the RAFAEL chatbot balanced an accessible interactive approach with medical safety, aiming to relay correct and verified information for the management of post-COVID-19. Development was followed by deployment with the establishment of partnerships and communication strategies in the French-speaking world. The use of the chatbot and the answers provided were continuously monitored by community moderators and health care professionals, creating a safe fallback for users. RESULTS: To date, the RAFAEL chatbot has had 30,488 interactions, with an 79.6% (6417/8061) matching rate and a 73.2% (n=1795) positive feedback rate out of the 2451 users who provided feedback. Overall, 5807 unique users interacted with the chatbot, with 5.1 interactions per user, on average, and 8061 stories triggered. The use of the RAFAEL chatbot and platform was additionally driven by the monthly thematic webinars as well as communication campaigns, with an average of 250 participants at each webinar. User queries included questions about post-COVID-19 symptoms (n=5612, 69.2%), of which fatigue was the most predominant query (n=1255, 22.4%) in symptoms-related stories. Additional queries included questions about consultations (n=598, 7.4%), treatment (n=527, 6.5%), and general information (n=510, 6.3%). CONCLUSIONS: The RAFAEL chatbot is, to the best of our knowledge, the first chatbot developed to address post-COVID-19 in children and adults. Its innovation lies in the use of a scalable tool to disseminate verified information in a time- and resource-limited environment. Additionally, the use of machine learning could help professionals gain knowledge about a new condition, while concomitantly addressing patients' concerns. Lessons learned from the RAFAEL chatbot will further encourage a participative approach to learning and could potentially be applied to other chronic conditions.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.036 | 0.070 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.002 | 0.001 |
| Science and technology studies | 0.003 | 0.002 |
| Scholarly communication | 0.004 | 0.003 |
| Open science | 0.002 | 0.005 |
| Research integrity | 0.002 | 0.003 |
| Insufficient payload (model declined to judge) | 0.006 | 0.002 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".