What do patients say and how do their opinions change? What messages do they convey about the quality of care and services? How do events like Covid affect their ratings?

The document “Using AI to understand the patient voice”* demonstrated the contribution of Artificial Intelligence (‘AI’) to understand the patient's opinion. Patients themselves can gain a different kind of insight through the comments they include in surveys about their care experience. When patients write comments, they describe all aspects of care that matter to them, whether they are the subject of a medical survey or not.

*https://catalyst.nejm.org/doi/full/10.1056/CAT.20.0103

AI can also automatically classify important topics (Figure 1) that can help providers to better respond to patients' changing needs.

AI categorises comments, sentences in the original text, so-called unstructured data.

The questionnaires can therefore evolve towards open questions and wider themes.

This same AI, being agnostic, can also be used to analyse the feedback from health professionals, concerning their expectations and appreciations.

Artificial Intelligence is a powerful tool for analysing these comments, to identify positive and negative trends (Figure 2) in patients' perceptions of care and services. For example, caring, kindness and empathy "in a stressful time" were recurrent aspects of positive care experiences.

At Phedone, our goal is to sort the most important from the rest.

Here are some results of the Phedone solution;

63% of automatically detected categories are industrialised in the support tools,
90% accuracy of automatic qualification,
Between 20 and 50% of staff time is saved.

We use AI, statistics and business rules to study the voice of the customer, the employees and now the patient and the medical staff in general.

Without any training phase required, the facility manager, the infrastructure and information system department and the quality department can immediately detect trends and weak signals. They can even sort and route support requests automatically.

Figure 1: trends examples

Figure 2; With a filter, the frequency and tone of comments

Phedone enables the translation of patient feedback into usable information. Phedone provides a toolkit to exploit the treasure trove you have in your current data. This "stuff" is often unused due to lack of resources. For example, the string diagram (Figure 3) immediately shows groups of individuals or functions that have made similar comments. The quality department or human resources department can set up working groups in a few minutes.

A white paper* and a dissertation** by David Sahuc draw a complete picture of the digital transformation of the healthcare system in the broadest sense, and its evolution towards more humanism thanks to technologies, liberating the time of professionals who will be able to better respond to patients' expectations.

*De la transformation numérique du systeme de Sante, David Sahuc, MBA IMT-BS

**Digital Transformation of the French Healthcare System, David Sahuc, MBA IMT-BS

“Health professionals spend a lot of time on activities that are not focused on their core business. The administration is overwhelmed by administrative formalities, including CERFAs, which require manual scanning, retention of paper documents, and document processing. 51% of hospital staff do not have enough time for all these tasks” David Sahuc.

Figure 3: Example of group, string diagram

In the quoted document, the patients distinguish between processes and people when describing negative experiences. Patients’ concerns and frustrations are largely driven by processes, not by the people involved in carrying out tests and treatments*.

*James, Tabitha L., Eduardo D. Villacis Calderon, and Deborah F. Cook. "Exploring patient perceptions of healthcare service quality through analysis of unstructured feedback." Expert Systems with Applications 71 (2017): 479-492.

Therefores, the comments need to be understood finely and also from a generic point of view.

Research shows some impact of patient feedback on medical practice and service in general but it is still too limited or non-existent*.

*Kumah, Emmanuel, Samuel E. Ankomah, and Felix O. Kesse. "The impact of patient feedback on clinical practice." British Journal of Hospital Medicine 79.12 (2018): 700-703.

Baines, R., Regan de Bere, S., Stevens, S. et al. The impact of patient feedback on the medical performance of qualified doctors: a systematic review. BMC Med Educ 18, 173 (2018).

Phedone allows you to open up open questions and not be afraid of overwhelming responses. Phedone works at the “Macro” level with trend extraction, but also at the “Micro” level, by tagging each verbatism with keywords (Figure 4).

Thus, each comment or support request is routed to the right person. Not only does the ticket not get lost in the support, but it is also resolved quickly. The support teams are relieved of their workload and can concentrate on other tasks. The author, whether a patient, doctor or employee is satisfied because the response time is drastically reduced.

“The aim is not only to give meaning to patient and nursing staff feedback, but also to encourage synergies between doctors, nurses and healthcare managers who are sometimes too closed off from innovation and progress. With Phedone, we detect the most original contributions , we notice the similarities of ideas and extract exploitable leverage of action”, Jonathan Foureur, CEO Phedone.

Figure 4: A verbatim, its keywords (here “velo”) and a sorting by opinion or originality index

In conclusion of the AI study*, the feeling of “compassion” for healthcare professionals is a strong positive signal in the patient's voice. We need to keep listening in order to learn more and adapt our actions more quickly, while showing professionals that their hard work int these challenging times is recognised with gratitude.

*https://catalyst.nejm.org/doi/full/10.1056/CAT.20.0103

In conclusion with Phedone, the verbatims collected are analysed very quickly with the creation of leverage that can be activated quickly. Messages are automatically routed and without waiting to be read. Staff are relieved of low value-added tasks.

Performance reminder:

63% of automatically detected categories are industrialised in the support tools,
90% accuracy of automatic qualification,
Between 20 and 50% of staff time is

Collective IntelligencePatient and medical voice

Discover the advanced AI thatautomates the processing of your customer returns

Get notified when we’re launching.