Gathering feedback from customers is fundamental for most companies – but few clients enjoy filling out long questionnaires. “Many people find it much easier to talk than to type,” says Romit Choudhury, Softbrik’s COO and co-founder. The company has therefore developed a tool that captures voice messages and turns the resulting audio streams into texts. Machine learning algorithms analyse the emotions transmitted by each respondent and detect key insights, which makes the evaluation of the feedback quick and efficient at scale.
Understanding patient feedback
Just as Softbrik had finalised a first version of its tool, intended for the telecommunications market, the COVID-19 pandemic broke out. The team realised that they could help doctors who were overloaded with patient calls. Supported by Luxembourg’s StartupVsCovid19 programme launched in spring 2020, it redesigned its tool for messages left by early-stage COVID patients, using artificial intelligence to help doctors detect the most urgent calls and identify patients speaking about similar symptoms. “This started our journey into understanding patient feedback,” recalls Mr Choudhury.
With this first experience in the healthcare field as a basis, the company saw that it could make a real difference in the field of clinical studies involving data from thousands of patients. “Traditionally, patients are asked to provide the initial batch of data during their first visit to the clinic, and then follow up at their next visit three months later or so. However, when people who have just learnt that they are severely ill are presented with a 30-page questionnaire, the quality of their answers will reflect their level of shock. Analysts have to deal with a lot of ‘data noise’.”
A complete digital journey
With Softbrik’s digital solution, patients will instead be asked to provide only the most essential input on an iPad or similar device via a mix of multiple selection questions, ratings, emojis and even voice messages. “The emojis work the best – people love them, and the smiling icons bring down their level of stress,” Mr Choudhury points out. A week later, when their shock has hopefully abated a little, they will receive an automatic notification on their phones to provide a next set of answers from the comfort of their home. Healthcare teams can also give patients a health card with a QR code to physically remind them to give timely feedback.
Our AI tools understand what the patients are talking about.
The digital journey will continue with regular, digital feedback provided by the patients on their state of health. Biostatisticians are supported by AI-powered dashboards. “Our AI tools understand what the patients are talking about – their reactions to the medicine, their sleep cycle, etc. – and can extract information about the context, their mood and so on.”
More data, less time and costs
The tool is already in use by pharma companies and hospitals in Switzerland, Germany and Belgium, and discussions with some US healthcare companies are under way. “Our clinical projects pass the ethical validation of the best pharmaceutical companies like Novartis, and we have proven that we can collect up to 25 times more patient intelligence while reducing the time needed for studies by 50-60% and budgets by millions of euros,” explains Mr Choudhury.
This is a unique advantage for companies targeting the international markets.
The focus is now on scaling and growing the team. The company is exploring cooperation opportunities with the University of Luxembourg and the Luxembourg Institute of Health, and Mr Choudhury is very active in the start-up community. “We are a close-knit group of founders, almost like Silicon Valley in the early 90s, and the government is extremely supportive. Our diversity is our strength: if I want to do a version of my tool in, say, Amharic, I can easily find 20 Ethiopian people in Luxembourg to test it. This is a unique advantage for companies targeting the international markets.”
Photo credits: Luxinnovation/Michel Brumat