It’s 2021, and the world is slowly leaving the COVID-19 pandemic behind. Not, however, without having acquired some new habits. For example, using apps for everything possible. During the pandemic, app usage has jumped by 40%, reaching an all-time high in spring 2020. Slight decrease since then isn’t significant enough to expect that it will all go back to the pre-pandemic levels. New habits, just like old habits, die hard. Apps are made to get you hooked. And this is especially true when it comes to healthcare apps.
For obvious reasons, the healthcare apps sector has benefited from the pandemic. Today, it attracts funding and investment like never before. According to a report from Mercom Capital Group, funding activity among digital health companies was up by 138% during the first half of 2021 compared to the first half of 2020. Raj Prabhu, CEO of Mercom Capital Group, announced that « the digital health sector had a spectacular first half of 2021. Venture investments in digital health during the first half of 2021 have already surpassed funding raised in all of 2020 and is the largest amount raised in a single year since 2010 ». mHealth apps alone got $1.6 billion. Among the ones to receive the most funding was Noom ― a weight loss app that plans to expand its influence to areas such as stress and anxiety, diabetes, hypertension and sleep. Among the funding champions we also got Kry, which allows the users to consult a qualified health professional from smartphone or tablet. Noom raised $540 million in funding, Kry got $300 million.
As healthcare apps are now successful and widespread, the companies behind them go through the common stages of competitor research, product development, reputation management, and customer care. Which means all of them are using (or should be using) sentiment analysis, which is beneficial for all of these stages. So let’s dive deeper and discover what sentiment analysis is, and how healthcare apps can benefit from introducing it into their practice.
What is sentiment analysis?
Sentiment analysis (also known as opinion mining and emotion AI) is a method of text analysis that identifies tones and emotions of written text. It’s a classic example of big data implementation. Sentiment analysis turns unstructured data into structured data that reveals meaningful, emotionally colored insights. It does this in four not-so-simple steps:
- Data collection
- Text preparation
- Sentiment detection
- Sentiment classification
The last step breaks down the data into positive, negative, and neutral pieces. This is when the user gets to see colorful graphs that tell them that, for example, their campaign was met positively by 53% of the ones who wrote about it, negatively by 7%, and the other 40% didn’t express either positive or negative sentiments.
The most popular tools with the sentiment analysis feature are social listening tools, like Brandwatch, Talkwalker, and Mention. Elinext has also developed its own sentiment analysis software: you can take a look at the software case study here. In addition to sentiment analysis, such tools often perform other types of analyses: growth, demographics, location, etc.
How can healthcare apps benefit from sentiment analysis?
- Competitor research
Competitor research comes before every app is developed, and continues throughout the app’s existence. Healthcare apps are not an exception. The app developers have to look closely at the healthcare niche they are interested in and research what is already happening there.
Let’s take Noom as an example ― the very successful weight loss app that we mentioned in the introduction. Weight loss is a very competitive niche, and before introducing yet another weight loss app the company had to find out what their competitors are doing wrong. For that, they could have used sentiment analysis among other tools. If they had applied sentiment analysis to most popular competitor brands (LoseIt!, MyFitnessPal, Weight Watchers App), they would have seen what the common target audience likes and dislikes about existing apps in this niche. Noom could have discovered that people hate calorie counting, and choose another way to manage weight and to promote the app. Noom was the first app to focus on the psychology of overeating, so we think this is exactly what happened.
Let’s take women’s health apps as another example. A quick sentiment analysis would quite certainly reveal that women are tired of pink flowery designs that almost every fertility or period tracker app has. A new app in this niche could get popular with simply a different design approach, and compete with the much-criticized pinky purple Flo, Natural Cycles, Eve, Period Tracker and Period Calendar, etc.
- Product development
While the product is being developed, it undergoes numerous analyses. Marketers and product managers gather feedback non-stop. Otherwise, they wouldn’t know which features are important, which were made badly, whether the app’s design is interactive and user-friendly, and whether the marketing tactic they’ve chosen works. Basically, they have to check what people like and dislike about the product all the time.
Predictably, this is when sentiment analysis enters the scene again. Sentiment analysis tools gather feedback on social media networks, blogs, forums, review sites, and bring back the results. This is much quicker than gathering focus groups, carrying out interviews, or persuading people to complete feedback questionnaires. And, once you have sentiment analysis software, the analysis happens all the time. You don’t need any more work or investment. It will show you straight away if your new design is disliked by the audience, or if people are unhappy with the new feature.
- Reputation management
Healthcare apps are popular and on the rise, but this doesn’t mean they don’t experience difficulties. One of the most apparent challenges is trustworthiness. Healthcare is a sensitive topic: people should trust the apps enough to share their private and often sensitive information with the app. This can be hard, considering that even real-life clinics with digital technology involved can be unsafe. Just recently, a finnish mental health startup Vastaamo experienced a catastrophic data breach. Turns out, the app had a security flaw in the IT systems, and the hackers were able to expose the entire patient database. The actual written notes that therapists had taken ended up in the hands of hackers who hunted down patients and asked each one for ransom, threatening to make their health issues, thoughts and feelings public.
When it comes to apps, the situation is even more alarming and unclear. Let’s take healthcare apps for women again: when Privacy International, a nonprofit group in Britain, tested 36 popular women’s health apps in 2018, they found that 61% automatically transfer data to Facebook. And this is sensitive data that concerns female health and topics such as periods, pregnancy, fertility, and birth control.
All in all, it’s easy to see why consumers might have trust issues when it comes to health apps. To battle this issue, healthcare apps can do about this:
- Secure data properly
- Have honest and transparent data policies
- Manage reputation
We won’t stop at the first two points: although they are definitely the most important ones, they are out of scope of this article. The author genuinely hopes you know that security has to be flawless and that lying to consumers is dangerous. But this is not enough. You also have to manage your reputation: know what the world is saying about you, which articles are being published, whether the app has been mentioned in a study similar to the one Privacy International has done, and what has been discovered. The bad word is not always true or fair, but it spreads fast.
Sentiment analysis tools do the following: they analyze the conversations around the brand in real time, and if there is a sudden increase in negative sentiment, they send notifications straight away. The marketers behind the healthcare app get such a notification, they can dig deeper, solve problems, reply to negative comments, fix what’s wrong, and generally manage the situation before it’s too late.
- Customer care
Healthcare apps need to take care of their customers just like other apps, if not more. The customer care or the customer support teams could benefit greatly from using sentiment analysis in their work. Primarily, as a way to catch negative comments, problems, and questions on social media networks and reply to them in time. These days people never call, rarely email, and only sometimes send private messages to companies on their social networks. Too often they just tag the brand (not always correctly) and wait for it to respond. The UK’s National Health Service did a whole study where they analysed the patient’s sentiment towards their hospitals. They wanted to capture patients’ experience and improve the hospitals’ services accordingly. Sentiment analysis allows companies to see all these mentions and show that they are the most caring healthcare app there is. Isn’t that a great way to stand out?
What are the problems and challenges of sentiment analysis?
Sentiment analysis is not without its problems, and, unfortunately, most of them concern its accuracy. Machines are yet to learn how humans really talk. Firstly, just like , machines struggle to understand sarcasm and idioms. Hence, if the analyzed piece of text is full of them, the results may simply be wrong. The second biggest problem is multilingual data. Right now, most sentiment analysis tools work best if data is in English and struggle with other languages. Machines learn, so this is likely to get better in the future. There are plenty of other minor challenges that affect the accuracy of the final results. However, with the right programming, most of them can be dealt with.
What is the future of sentiment analysis?
Sentiment analysis has come a long way since it was first introduced 20 years ago. But there is still more to develop. In the future, sentiment analysis is expected to become more accurate and more detailed. New versions of sentiment analysis, such as voice sentiment analysis and video sentiment analysis, are expected to enter the market. Analyzing voice and videos will have new implications for healthcare apps, the examples of which we’ll be excited to see.
Final words
For healthcare apps, the stakes are often higher. Every complaint might become something huge ― something that would lead to a scandal and ruin their reputation. Every development problem might have serious consequences when the niche is as serious as that one. This is when only the implementation of big data can help. Sentiment analysis is like a spy that keeps an eye on everything that might signal a problem and delivers the colorful reports to make sure you know.
