How we improved online therapy platform performance with matching algorithms

Over the past few years, online therapy platforms have seen a surge in demand; with COVID-19 locking people inside their homes, that trend only amplified. 

They offer comfort, convenience, and privacy to those who are looking for emotional support. While they are indeed highly convenient, their concerns stem from matchmaking accuracy to therapists feedback evaluations, quality sessions, and the most critical one, user engagement with retention rate.

Many online therapy services also share the challenges that picking a matching algorithm can solve.

Their accuracy and efficiency are further refined using machine learning techniques, with the help of earlier matches and client satisfaction. 

However, for this article, we shall only speak about the classic string-matching algorithm—the same as those used in dating apps.

We will also provide an illustrative case of how these algorithms can correctly match clients with therapists on this basis. Well, let’s get started and explain how it all works!

Online therapy platforms use algorithms to match patients with the best therapist based many data points. There are some interesting algorithms that have been designed to analyze detailed user profile information, containing things like mental health problems of the patient, therapy goals, personality traits and preferences.

These algorithms are built upon psychometric assessments, such as questionnaires that highlight elements for a successful therapeutic session. This data help the algorithm to match patients based on their requirements and therapy type goals with therapists.

The algorithms would also consider factors beyond these assessments (e.g., the therapist's concentration, availability, and even geographical proximity if in-person sessions were preferred).

While the exact algorithms are often kept private, they aim to make therapy even more effective by carefully matching patients and therapists. 

They continuously measure and learn from the performance output of these algorithms to ensure that their efficacy evolves over time as part of online therapy development medical software development.

As we said earlier, one of the primary challenges in online therapy is the question - to which therapists do you match your clients? 

This is tough, as we know many things come into play here. Availability, expertise and of course personality/communication style, fees, etc. Despite working through various different therapists over time, many patients eventually just drop out of psychotherapy because they cannot seem to get connected.

Another major problem is the lack of efficiency in manual processes. Other platforms can take a week and as long as 7-15 days since some therapists still compare data manually, make transactions, or even set up schedules. 

It is very easy to lose vital information where manual processes have plenty of room for error - losing important notes during a therapy session, failing to track the progress or follow up on sessions ultimately derailing your therapeutic journey.

Matching algorithms are therefore fundamentally concerned with the capability to analyse data, and from that analysis they match coherent outcomes across a multitude of use cases. In the mental health space, effective patient matching algorithms can drastically reduce wait times and much improve client satisfaction and therapy outcomes.

Matching algorithms provide a strong workaround by analyzing client profiles, preferences and feedback to help find the right therapist fit. 

This particular method has worked well in the industry and can be seen from one of our own clients. Here, we’ll discuss how this strategy proved successful and its resultant effects.

Meela Health is a platform focused on offering better psychotherapy for everyone. Although it is mainly marketed towards women and to the Swedish market, it does not discriminate against any other gender and offers the same service to everyone. 

The reasoning behind being primarily marketed towards women is the fact that women are a lot more likely to seek psychotherapy compared to men and others (1 in 4 patients attending psychotherapy are men).

The main idea behind Meela was to provide the patients with a platform that utilizes a patient matching algorithm to ensure they get better psychotherapy and, at the same time, provide therapists with patients they will have more success with. 

The result would be a higher likelihood of patients staying in therapy and getting the help they need to resolve their issues. 

The issue Meela identified was that too many patients will give up on psychotherapy due to not being able to find the right therapist for them, even after trying multiple different therapists. 

This, in addition to just making the decision to try psychotherapy, would create unnecessary friction and put patients in very uncomfortable and hopeless situation which would result in them giving up on psychotherapy entirely. 

To address this issue, one of Meela’s co-founders, Tiffany decided to try and create a questionnaire that would be used to match patients to therapists. 

This was possible due to Tiffany being educated in the psychology field as well and the questionnaires would be filled out by both the therapists and the patients. 

To achieve good results all the questions from the patient’s questionnaire would have the matching question in the therapist questionnaire.

However, going through all of these questionnaires manually for each and every patient and then going through all of the therapist questionnaires to find the right match was a lot of manual work and pretty inefficient. 

It wasn’t for nothing though, as using this method was how Tiffany proved that their method worked and created, what I would consider, a true MVP (minimum viable product) in every sense of that term. 

Once the prototype was tested, automating the process and creating an application that would offer this service was the obvious next step.

The key issues the application had to resolve:

• Matching algorithm
• On-boarding for patients and therapists
• Human evaluation of results
• Confirmation and securing of identity

Meela engaged with us to help them build their tech. We were key players in solutions that identified and addressed each of these shortcomings. 

Next, let's get to the details of how we overcame these challenges and made Meela a more effective online therapy service provider.

Matching algorithm was the next step in the process of matching the patients to therapists and the main feature of the Meela app. It would remove the burden of manually going through each and every patient questionnaire and then doing the matching. 

To make sure that the algorithm worked correctly and provided quality matches, each question was graded in importance along with the answers. 

There were also “critical” question and answer pairs that would eliminate the therapist from matching with the said patient entirely. 

Some examples would be if the patient and therapist don’t speak the same language or if the patient wants in-person therapy, but the therapist was too far away from the patient. 

It would also handle more specific psychotherapy cases where it would eliminate therapists that strictly said they don’t want to work with the problem area the patient was suffering from; and these problem areas and expertise were the main matching criteria, ensuring the patients would get the highest chance of success in therapy. 

Given the sensitive nature of therapy, the algorithm was manually reviewed by Tiffany to ensure it aligned with best practices. While the algorithm was strong, it needed a few tweaks and fine-tuning to make sure it provided the highest quality of matching. 

It would also take into consideration the availability of therapists (how many patients were already matched to the therapist) ensuring that the patients wouldn’t have to wait for a long time to start their therapy.

Patient and therapist onboarding was the main point of friction for the application, as the users would be greeted with a lengthy questionnaire as soon as they would sign up. However, this was a necessary step to ensure that the application can provide quality matches. 

The main problem was designing and implementing a good user experience that would guide the user through the questionnaire and make it as easy as possible for them to go through the process with the least amount of friction. 

This consisted of having multi-step forms with various input fields and complex validation logic in each step, along with detailed tooltips to guide the user through the entire process. 

This entire process went through a lot of iterations based on user testing and changes to the underlying questionnaires and was reworked a lot throughout the development and design process of the application. 

This was very necessary, as it is super important to identify critical friction points for users which would result in user churn and the onboarding process was the most critical point for this specific case.

Human evaluation was determined as a critical part of the system from the very start. Due to the sensitive nature of the problems the application tries to solve, it was absolutely crucial to have an additional step of human evaluation before the patients would receive the results of the matching algorithm. 

This process is recommended in any software that has automation steps and handles any sensitive topics like health, especially so in the case of Meela which handles psychological health areas. 

The way this worked was that the matching algorithm would be in charge of determining the best 10 matches for the given patient, the data for the patient along with the data about 10 best matching therapists would then be stored and viewable in the admin panel. 

Tiffany would then manually review this data, gaining access to the question and answer pairings, matching scores, and the 3 suggested therapists for the patient in issue. 

Following a thorough examination, she would validate the suggested therapists and provide the patient with contact details along with information regarding the following steps. 

This procedure was important in the evaluation of the matching algorithm and in ensuring the system’s integrity.

Although the number of cases that needed a manual check at first was considerably high, it reduced as the algorithm was refined in the future. However, all these advancements make it imperative that even though it may take a shorter duration the work calls for human input.

Identity confirmation and protection is a necessity in any system working with sensitive personal information. It is absolutely crucial to ensure that the information shared by the user is stored securely and protected from any outside viewing. 

In the case of Meela it was also very important to ensure that the users accessing the application were real, verified people so that both the patients and the therapists could be confident that they are interacting with other real human beings. 

This would give users more confidence in the system and also protect the system from fraudulent activity, as the identities of the users were known and it created an atmosphere of trust. 

Ensuring that the identities of the users were known and valid was done with BankID integration. BankID is a popular method of confirming identities of people in Sweden and it is almost as valid as an actual country issued identification. 

In such systems it is very important to have these types of measures in place to ensure the users feel more comfortable and confident when signing up to the system.

Other than identity verification, the application also ensured the privacy of user data, obscuring and hiding any personal information until it was absolutely necessary to share basic information which was in the last step of the process. 

This last step of the process was the step that would connect the patients to the recommended therapists and in this step the therapist information would be shared with the patient, and once the patient would select a therapist, the patient’s information would also be shared with the selected therapist. 

The data was however obscured and protected in the database and stored securely.

No matter whether you have a set of questions on paper or if you have been dealing with large-scale work, which means numerous questionnaires from patients and therapists in the field, organizing their work through each completed document can be very time-consuming and inefficient.

Nevertheless, this approach was vital for Tiffany to show the efficiency of their way of working, also, it is possible to conclude that they created what can be defined as the most minimally viable product.  

When the prototype was verified, it only made sense to take the next step and create an application to put this service into practice.

Therefore, it can be stated that matching algorithms can greatly improve the efficacy and efficiency of the online therapy platform. 

Through use of big data and predictive analysis, apps such as the Meela Health are capable of providing better pairings of the patient and therapist.

Their proposed solution is 25× as efficient as other providers so that clients build a connection with their therapists. 

This specific focus in care not only aims at refining the clinical consequence but at the same time at a higher degree, the quality of life of those patients and individuals we have working in our organizations.

Given that online therapy is still a relatively young field, the focus must always be on working on and improving the matching algorithm in order to level up with other platforms and provide the best experience for the user.

If you want to learn more about how matching algorithms can advance your online therapy marketplace or better your healthcare application, then we would like to hear from you.  

Whether you already have a clear idea as to how you want the system to work or have already developed a concept but need help fine tuning it we are able to assist you at any point of the process.

Why not grab a quick e-coffee with our growth manager Haris for the possibilities of setting this new model? 

It is a very useful idea to work on the possible problem-solving aspects and know how we can be of help to you. Please click here to schedule a time with Haris.

In collaboration, let us turn the screen into reality for a positive change in the impact of healthcare facilities.