The emergence of evidence-based medicine in the early 1990s soon led to efforts in psychotherapy to prove the effectiveness of existing interventions beyond doubt. This has undoubtedly been achieved: common psychotherapeutic methods are effective in treating almost all mental disorders. Statements about the evidence base of psychotherapeutic methods are based on studies conducted under highly controlled conditions. These are necessary to determine the effectiveness of an intervention, but often do not reflect everyday therapeutic practice. This is one of the reasons for the research-practitioner gap: the evidence-based practices prescribed by efficacy research are rarely used by psychotherapists in practice because they are perceived as too inflexible and do not fit in with their own working style and clinical reality.

Decades of research into effectiveness have therefore produced a large number of manualized psychotherapeutic procedures for treating specific disorders, whose effectiveness is well documented but which can hardly be personalized.

There is little difference in effectiveness; most patients benefit from all of them to roughly the same extent. This proliferation of methods ensures a wide range of options, but it is still unclear which patients benefit most from which methods: the evidence of effectiveness applies only to the “average patient.” But even here, 30% of patients do not benefit from their treatment, and in about 10% of cases, the burden increases during therapy. In addition, there are premature discontinuations of treatment before the desired success is achieved.

These problems suggest that existing evidence-based psychotherapies should be implemented more effectively and further optimized. In order to improve psychotherapeutic care for people with mental illnesses, practice-oriented research is an important complement to evidence-based practice.

Treatment guidelines are not derived from controlled studies. Instead, practice-oriented research draws its data directly from everyday clinical practice. This allows systematic information to be obtained that can support clinical decisions. Psychology can draw on a long tradition here: as early as 1954, Paul Mehl presented a convincing paper showing that medical and psychotherapeutic practice can be significantly improved when supported by data-based decision-making aids. Practice-based research has begun to implement Mehl's ideas and has been continuing for several years under the name precision mental health. Improvements in statistical methodology and easier data collection and evaluation thanks to technical innovations have further advanced practice-oriented research.

The questions left unanswered by evidence-based psychotherapy should be answered. Their contributions have led to the following achievements:

• The best possible intervention for the individual patient can be found using statistical algorithms that work with data from routine care.
• Early detection of unfavorable therapy outcomes can be ensured with the help of monitoring systems. In the event of negative developments, therapeutic techniques that have been helpful in similar situations can be recommended.
• Instead of planning an entire psychotherapy session in advance, adaptive decision rules can suggest the appropriate intervention to therapists during ongoing therapy.

The University of Greifswald is also working on expanding and testing these methods. We have formed a practice-research network to implement these research approaches. It consists of the Schön Psychosomatic Clinics in Germany, Philipps University of Marburg, and the University of Greifswald, as well as the Center for Psychological Psychotherapy (ZPP) located there.

Within the framework of this Germany-wide practice-research network, we are investigating inpatient and outpatient psychotherapy concepts using various statistical methods in order to identify starting points for optimization and enable more effective implementation of evidence-based psychotherapies.

These include both data-driven, theoretically agnostic data analysis methods from machine learning algorithms and theory-driven models that mathematically specify mechanistically interpretable relationships between variables (often including both observable variables and postulated, theoretically meaningful hidden variables, e.g., through Bayesian statistics). In addition, we are investigating whether computer-assisted evaluation of video-recorded therapy sessions can be used to make statements about the success of therapy. This involves the use of motion energy analysis (MEA) techniques, which can be used, for example, to measure the synchrony of the movements of the therapist and patient. In recent pilot studies, this physical synchrony has proven to be a strong predictor of therapeutic success and a correlate of a positive therapeutic relationship.

At the heart of our practice-oriented research is the “Greifswald Psychotherapy Navigator System (GPNS)”, which is designed to provide our psychotherapists with concrete and, above all, evidence-based recommendations for selecting and adapting their psychotherapeutic approach.