Trial document




drksid header

  DRKS00011555

Trial Description

start of 1:1-Block title

Title

Early detection of adverse outcomes in Dialectical Behavior Therapy with Machine Learning

end of 1:1-Block title
start of 1:1-Block acronym

Trial Acronym

[---]*

end of 1:1-Block acronym
start of 1:1-Block url

URL of the Trial

[---]*

end of 1:1-Block url
start of 1:1-Block public summary

Brief Summary in Lay Language

Borderline personality disorder is characterized by symptoms such as emotional dysregulation, self-harm, feelings of emptiness, identity disturbances, or suicidal ideation. Psychotherapy can alleviate these symptoms and help to have a fulfilling life. Studies show that Dialectical Behavior Therapy (DBT) is an efficacious treatment for borderline personality disorder. In DBT, patients learn to cope with the ups and downs of their feelings. In the long run, this helps to alleviate symptoms in most patients.

Unfortunately, up to one third of all patients drop out of DBT. This study aims to identify possible premature treatment terminations before they occur, in order to allow more patients the optimal treatment effect.

There are many reasons for premature treatment termination. Most studies have assessed these reasons only once at the start of the treatment. This means that changes happening during the course of the therapy are not assessed.

In this study, an early-warning system for premature treatment termination will be developed. Borderline patients receiving DBT will speak or type one short text daily about how they think about therapy and everyday live. Their answers will be collected with an app and analysed with speech recognition programs and machine learning systems. Machine learning can be used in order to learn which combination of words can predict premature treatment termination. The early-warning system should be able to identify complex speech patterns, which are associated with possible premature termination of treatment. In the future, this could help to adapt the therapeutic intervention, if a patient wants to drop out of treatment or cannot benefit from DBT.

end of 1:1-Block public summary
start of 1:1-Block scientific synopsis

Brief Summary in Scientific Language

Borderline personality disorder (BPD) is characterized by severe psychological distress as well as high health care usage. BPD involves severe affective dysregulation, identity disturbances, and social problems. Typical behavioral problems are self-harm, suicide attempts, impulsivity as well as depressive symptomatology.

Dialectical Behavior Therapy (DBT) is an evidence based psychotherapy of BPD. A meta-analysis of 28 studies showed that reductions in general psychopathological symptoms, anger as well as parasuicidal behavior were significantly higher as compared to treatment as usual (Stoffers, Völlm, Rücker, Timmer, Huband & Lieb. Cochrane Database Syst Rev, 2012.) Despite these overall promising results, premature treatment termination is a problem in DBT and varies between 10 and 30 % (Bohus et al. Behav Res Ther. 2004 May;42(5):487-99. Stiglmayr et al. Borderline Personal Disord Emot Dysregul. 2014 Dec 18;1:20.) Identifying risk factors and predictors of premature treatment termination is a core challenge.

In the literature, different predictors of premature treatment termination have been identified and were usually measured with single baseline assessment. This method does not investigate dynamic processes. Because of the usage of a predefined set of predictors, it remains an open question which other factors might have an impact on adverse treatment outcome.

Machine learning (ML) technology enables identification of parameters which predict premature treatment termination without having to predefine predictors.

In this study, predictors of premature treatment termination in DBT for BPD will be identified. Speech recognition and ML technology will be used to analyze comments of patients about the therapy and their everyday lives in a specifically designed app. Additional standardized questionnaires will be assessed. In the long run, the ML system will be used for early detection of possible premature treatment terminations, in order to adapt the therapy accordingly.

end of 1:1-Block scientific synopsis
start of 1:1-Block organizational data

Organizational Data

  •   DRKS00011555
  •   2017/03/30
  •   [---]*
  •   no
  •   Approved
  •   2017-511N-MA, Medizinische Ethik-Kommission II Medizinische Fakultät Mannheim der Universität Heidelberg
end of 1:1-Block organizational data
start of 1:n-Block secondary IDs

Secondary IDs

  •   U1111-1193-1261 
end of 1:n-Block secondary IDs
start of 1:N-Block indications

Health Condition or Problem studied

  •   F60.31 -  [generalization F60.3: Emotionally unstable personality disorder]
  •   F60.30 -  [generalization F60.3: Emotionally unstable personality disorder]
  •   F91 -  Conduct disorders
  •   F43.1 -  Post-traumatic stress disorder
end of 1:N-Block indications
start of 1:N-Block interventions

Interventions/Observational Groups

  •   Patients receiving residential or outpatient DBT answer question in a therapy app on a daily basis and additionally fill out standardized online questionnaires on a monthly basis.
end of 1:N-Block interventions
start of 1:1-Block design

Characteristics

  •   Non-interventional
  •   Observational study
  •   Single arm study
  •   Open (masking not used)
  •   [---]*
  •   Uncontrolled/Single arm
  •   Treatment
  •   Single (group)
  •   I
  •   N/A
end of 1:1-Block design
start of 1:1-Block primary endpoint

Primary Outcome

1. Premature treatment termination (yes/no) - information supplied by the therapist if patient has terminated the treatment prematurely. A time point for this assessment cannot be pre-defined. This variable is assessed only of a premature treatment termination takes place. It is assessed at the time of the termination.

2. Item: "Urge to terminate treatment" (visual analogue scale from 0-10) - daily rating by the patient in the app


end of 1:1-Block primary endpoint
start of 1:1-Block secondary endpoint

Secondary Outcome

Self-harm (frequency/day) - daily rating by patient in app

end of 1:1-Block secondary endpoint
start of 1:n-Block recruitment countries

Countries of Recruitment

  •   Germany
end of 1:n-Block recruitment countries
start of 1:n-Block recruitment locations

Locations of Recruitment

  • Medical Center 
end of 1:n-Block recruitment locations
start of 1:1-Block recruitment

Recruitment

  •   Actual
  •   2017/06/22
  •   100
  •   Monocenter trial
  •   National
end of 1:1-Block recruitment
start of 1:1-Block inclusion criteria

Inclusion Criteria

  •   Both, male and female
  •   16   Years
  •   no maximum age
end of 1:1-Block inclusion criteria
start of 1:1-Block inclusion criteria add

Additional Inclusion Criteria

Emotionally unstable personality disorder (F60.30 or F60.31)
Posttraumatic Stress Disorder (F43.1) or
Conduct Disorder (F91)

end of 1:1-Block inclusion criteria add
start of 1:1-Block exclusion criteria

Exclusion Criteria

Alcohol dependence,
Substance dependence,
Acute psychosis

end of 1:1-Block exclusion criteria
start of 1:n-Block addresses

Addresses

  • start of 1:1-Block address primary-sponsor
    • Zentralinstitut für Seelische Gesundheit
    • Mr.  Prof. Dr.  Martin  Bohus 
    • J5
    • 68159  Mannheim
    • Germany
    end of 1:1-Block address primary-sponsor
    start of 1:1-Block address contact primary-sponsor
    end of 1:1-Block address contact primary-sponsor
  • start of 1:1-Block address scientific-contact
    • Hochschule Darmstadt, Institut für angewandte Informatik
    • Mr.  Prof. Dr.  Bernhard  Humm 
    • Schöfferstr. 8b
    • D-64295  Darmstadt
    • Germany
    end of 1:1-Block address scientific-contact
    start of 1:1-Block address contact scientific-contact
    end of 1:1-Block address contact scientific-contact
  • start of 1:1-Block address public-contact
    • Hochschule DarmstadtInstitut für angewandte Informatik
    • Ms.  Dipl.-Psych.  Nora  Görg 
    • Schöfferstr. 8b
    • 64295  Darmstadt
    • Germany
    end of 1:1-Block address public-contact
    start of 1:1-Block address contact public-contact
    end of 1:1-Block address contact public-contact
end of 1:n-Block addresses
start of 1:n-Block material support

Sources of Monetary or Material Support

  • start of 1:1-Block address materialSupport
    • Hessisches Ministerium für Wissenschaft und Kunst
    • Rheinstraße 23-25
    • 65185  Wiesbaden
    • Germany
    end of 1:1-Block address materialSupport
    start of 1:1-Block address contact materialSupport
    end of 1:1-Block address contact materialSupport
end of 1:n-Block material support
start of 1:1-Block state

Status

  •   Recruiting ongoing
  •   [---]*
end of 1:1-Block state
start of 1:n-Block publications

Trial Publications, Results and other Documents

  • [---]*
end of 1:n-Block publications
* This entry means the parameter is not applicable or has not been set.