# Book - Digital Therapeutics for Mental Health and Addiction. The State of the Science and Vision for the Future Edited by Nicholas C. Jacobson Center for Technology and Behavioral Health, Geisel School of Medicine at Dartmouth College, Lebanon, New Hampshire, United States; Department of Biomedical Data Science, Geisel School of Medicine at Dartmouth College, Lebanon, New Hampshire, United States; Department of Psychiatry, Geisel School of Medicine at Dartmouth College, Lebanon, New Hampshire, United States; Department of Computer Science, Dartmouth College, Hanover, New Hampshire, United States Tobias Kowatsch Future Health Technologies Programme, Campus for Research Excellence and Technological Enterprise, Singapore-ETH Centre, Singapore, Singapore; Centre for Digital Health Interventions, Department of Management, Technology, and Economics, ETH Zurich, Zurich, Switzerland; Centre for Digital Health Interventions, Institute of Technology Management, University of St. Gallen, St. Gallen, Switzerland; Institute for Implementation Science in Health Care, University of Zurich, Zurich, Switzerland; School of Medicine, University of St Gallen, St Gallen, Switzerland Lisa A. Marsch Center for Technology and Behavioral Health, Geisel School of Medicine at Dartmouth College, Lebanon, New Hampshire, United States; Department of Biomedical Data Science, Geisel School of Medicine at Dartmouth College, Lebanon, New Hampshire, United States; Department of Psychiatry, Geisel School of Medicine at Dartmouth College, Lebanon, New Hampshire, United States # Chapter 2 - Mental Health Disorder Prevalence This chapter talks about the prevalence of mental illness 2.9 - Barriers for traditional healthcare, thus leverage digital therapeutics Blended approach: Free up time of clinicians to help more people # Chapter 3 - 3.1 Internet-Based Programs: Substance Abuse Barriers = Cost, Don't know where to go Evidence supporting the efficacy is the same as face-to-face in delivering care. Components usually = MI Skills, iCBT, home work assignments. Can increase accessibility, hard to reach population. 3.1.1 - Internet Based Programe vs Mobile App - Not dependent on Mobile OS, - Cheaper - But less access to sensors 3.2 IBP for Substance Use Disorder - CBT, MI, Mainly - The limitation is that it's in Research. Sampling could be biased as well. / Representative 3.2.1 Therapeutic Education System (TES) - C.T Component - Psychoeducation - Lessons - Rewards could be encouraging messages or money to enforce behaviour. - Several research showed promise in TES in Drug use. - Market in the digital therapeutics as prescribed intervention, - ACHESS - Is another app, based on self-determination theory for substance abuse - ==Need to be based on theory== # IBP for other illnesses 3.3 Mindfulness, CBT, psychotherapy - Apps for other disorder 3.3.1 - Mood and Anxiety Not enough service providers Self-help - referral - clinicians (page 23) ==stepped care approach== CBT - 8 Sessions - Guided by staff For depression - Monitor/ assess safety (suicidality), referral for additional treatment. Page 24 - Conflicting adherence Rates reported. ## Adherence Measures - Page 25 - Define Adherence Rate (Kelsders et all, 2012) Low use doesn't mean no use. - No consistent measure - Selection bias of sampling - Blended approach: human engagement factors increase engagement and increase more use. iCBT is more used. - There are many IBPs in the market, hard to compare due to inconsistent measures and heterogeneity of features; sampling: more need to be done.. need to have control group, and RCT. # Chapter 4 (Page 31) Second Wave are tools with the aim of helping people manage mood and behavior (mostly apps). (First wave are Internet based for addiction, ) - Don't know which type. more effective - Consumer app vs Clinical Grade (Research Grade) - 10-20k apps on the app store -- coach guided, self-guided. - Mindfulness, online peer community, games, tracking. - What are the evidence? - Different types: - ## Page 32 - Guided digital therapeutics ^5a605b - Apps that have educational components - Apps activities - for habit control, mood management - Support from clinics / trained coaches are critical - Based on CBT, dialectic therapy, problem-solving treatment, behavioural activation, and exposure therapy. - ==More effective to have clinical support then patients using it on their own (Wright et al, 2019)== # HOPES (1) Triage (2) Detection/Intervention (3) Treatment (Blended) - Need to be clear if we are doing all these. - Talkspace / Betterhelp - The access to clinicians via tele-consult on demand is evidence-based. Unsure Having the ability to send/receive messages from clinicians makes any difference. - Page 33 - Evidence base for Guided digital therapeutics - Most evidence show short term effect - =="Self guided tools exists but are meant to be used in conjunction with face to face treatment (Owen et al), encourage blended approach, face to face context." - PTSD There are some apps that are not integrated to clinical care. - Page 33 - Self -guided second wave Digital therapeutics - - App base on evidence based exercises without coaches/clinicans - ==Self guided digital therapeutics are not as effective for moderate to severe mental and behavioral health problems, but do have an effect on mild mental health challenges -- a list of citation==. ((Arean et al., 2016; Cuijpers et al., 2011; Drissi, Ouhbi, Janati Idrissi, & Ghogho, 2020; Pratap et al., 2018;Rathbone&Prescott,2017)) - Page 34 - Usually based on CBT, problem-solving therapy, dialectual, ACT with features such as the above, activity, goal-setting, tracking, behavioral, mood tracking ,education about coping methods. - Becausee there are consumer grade, can't control how it should be used. - User use for a while then stop. - ==Engagement could be a factor, compared to guided digital therapeutics (mohret et al 2019) Evidence Base for self-guided digital therapeutics - Not much evidence loneliness - compare no treatment, better (park et al 2020) p34 - Mood Tracking app - ==Scheer & Goerng 2020, mood tracking itself can lead to helpful information seeking for mood disorder. - Commonly use with EMA - - Based on EMA to recommend activity, resources, Examples of these apps 4.2.1 Evidence base for mood tracking app (P35) - Depress, anxiety - Few trials measure outcome because only tracking, mostly look at usability, feasibility, Mindfulness App - Cite benefits and sources - Specific for depression, anxiety -- Headspace. - Evidence based - Evidence of mindfulness on depression/anxiety - Caution of heterogeneity Online Peer Community (Page 35) - Research for that - Usually appear in existing social and networking sites like Facebook. 7 Cups 4.2.2 Evidence of peer community - effectiveness in depression - clinical outcome - citations, for stigma, " i feel connected" Serious Games - Cognition performance and mood - Novel for treating depression. - Some have gained FDA approval to target cognitions Evidence for serious games - 4.3 - Cautions and limitations with second gen digital therapeutics ==Engagement - Long term engagement is peer. Clinical engagement Research engagement is higher then real world because of incentive Less then 1% of consumer complete all sessions Associations between engagement and outcome not clear 2 week seems to be long enough for the effects, and user cycle back to using them when there is a need.== ==Some barriers to consider== 4.5 There is no clear guidance on how to use these tools in clinical setting ==p 39 - If the data are not integrated to medical records, the clinicians will need to spend additional time looking at them much like paper and pen. ==p39 - The therapy recommend in the app should be what the therapist, clinicans have been using, if not it will be confusing to both patient and doctors. # Chapter 5 : Blending digital therapeutics with Health Care ^924624 Page 47 - [[The definitions of Blended Care]] ==Added value of blended care. Pros== Page 46 - Blended Care Terminology 5.2.2 - [[Blended Care should be understood as a continuum]] Integration of technology with clinicans' work ==step care model, depending on the patient's need== Page 48 - Therapeutic support --> but how much is enough? What is optimal? Who should provide the coaching? Some say no difference, depending on patient's need. 5.3.2 - Digital components ==Page 48 - Provider need to constantly talk to patients to check in with their experience with tech/app -- to encourage regular use. ^e1f489 Page 48 - Ideally, should customise and select features that will be useful for patient, clinicians should be familiar with the features. 5.3 - Depending on the phase of recovery to use certain feature more. P49 - Blended Care approach - Why CBT so commonly use in blended care? Reasons - Acceptance and Commitment Therapy is possible also. Page 50 - Transdiagnostic Approach. Why? Features 5.5 - page 50 - Common Features of Blended Care Treatments - What has been done.. Current commercial grade not enough evidence. 5.5.2 - Value of Blended Care - ==Can cite the evidence here for value of blended care== 5.6 Illustrated, explain of depression Blended Care approach - RCT of Hellobetter Courses - Support by psychologist. - - The rest of the chapter offer examples of Blended Care, how apps/services have been adopted. ==we want to create a blended care, step wise interventions, but we are also buidling interventions/features ourselves and integrating them into our current model of care. The challenge is deciding features, and how to integrate blended care with the rest of the MDT memebers -- how to convert blended care/step wise for the rest of the MDT specialist? For example. Family Therapy: - Blended Care for CM, - Blended Care for MSW - Blended Care for OT - Blended Care for psychologist . Digital therapeutic guided - components features ![[Screenshot 2023-08-22 at 3.27.53 PM.png]] # Chapter 6 - Receptivity to Mobile Health Interventions - Patient's readiness to using the apps features, or EMI. - Ideally, it should be "Just in Time" - page 66, 67 - How to detect what is the right time to provide EMI? - There is a table in page 71 that summarises the factors signal that can be use to detect/predict receptivity - ![[Screenshot 2023-08-22 at 3.30.50 PM.png]] # Chapter 7 - Adapting just-in-time interventions to vulnerability and receptivity: Conceptual and methodological considerations Page 78 - Essentially, how to create Just in Time Adaptive Intervention? For example, in page 78 - "Every 2 hours, IF negative affect = Yes or presence of other smokers = Yes; and driving = No THEN, intervention option = Deliver a prompt recommending self-regulatory strategies ELSE, intervention option = No prompt" - Individual components, how do we decide which parts are important? like "Negative effect", or "Presence of other smokers" why 2 hours? - - We can only know about this by collecting long-term data, (intensive longitudinal data) and looking for correlations and create a model to test out hypothesis. ![[Screenshot 2023-08-22 at 3.33.49 PM.png]] Page 81 there is a table, showing how to operationalise scientific questions, and the methods used that provide answers. ![[Screenshot 2023-08-22 at 3.35.35 PM.png]] # Chapter 8 - A digital therapeutic alliance in digital mental health - Page 87 - 8.2 -- Therapeutic Alliance (Citations here) explains the importance of therapeutic alliance in treatment. - Applying Therapeutic alliance to Digital tools - do patients feel therapeutic alliance to the tools? -How to define this 8.4 - ![[Screenshot 2023-08-22 at 3.40.54 PM.png]] ==personally, i think that if the tool can be more personal with artificial intelligence, perhaps there will be more rapport. I am not sure if people can form rapport with just a features== # Chapter 9 - Conversational agents on smartphones and the web # Chapter 10 -Voice-based conversational agents for sensing and support: Examples from academia and industry - Page 119 - ==There are some prototypes used in diabetes, depression, mood assessment using voice-based conversational. - 10.3.2 - 10.3.2.1 Automated assessment of mental health disorders As an alternative to active sensing, voice assistants could potentially sense passively, that is use speech data to measure vocal biomarkers and detect or predict a state of vulnerability. Although, to the best of the authors’ knowledge, there is noresearch on VCA for speech-based automated mental health or drug use assessment, there has been extensive research on audio and verbal features to identify such disorders (i.e., vocal biomarkers). 10.3.2.1.1 Biomarkers for depression and suicidality Cummins and colleagues (Cummins et al., 2015) reviewed studies analyzing speech to diagnose depression and suicidality. The authors found prosodic and acoustic features to be associated with depression and suicidality. In particular, prosodic features, such as a reduction of the fundamental frequency variation (i.e., the rate of vibration of the vocal cords), energy, and speaking rate have been observed to be more prominent in depressed individuals. Acoustic features reflecting the airflow through the vocal cords, such as jitter (i.e., the variation of frequency between cycles of opening/closure of the glottis), shimmer (i.e., the variation of amplitude between cycles), and harmonic-to-noise ratio (i.e., a measure of the relative noise in the voice) have also been observed to correlate with depression. These features reflect the general tendency of depressed individuals to present a reduction in movement of the vocal fold and, thus, in speaking effort. According to the authors, however, the acoustic features have been proven to work on held vowels tasks but not on continuous speech and may not be appropriate for passive sensing through a VCA, which generally involves short commands or sentences. 10.3.2.1.2 Biomarkers for psychiatric disorders Later, Low and colleagues (Low et al., 2020) performed a similar systematic review for automated assessment of psychiatric disorders (i.e., depression, post-traumatic stress disorder, schizophrenia, anxiety, bipolar disorder, bulimia, anorexia, and obsessive-compulsive disorder). Almost half of the studies investigated acoustic features as biomarkers for depression. Others mainly included schizophrenia, bipolar, and post-traumatic disorder. Moreover, although acoustic features such as jitter and shimmer significantly correlated with the frequency or severity of both depression and anxiety, this latter represented a mere 5% of the studies. Moreover, the authors reported which devices were used to collect such types of data (see https://tinyurl.com/tu58te3) and observed that the reviewed studies included telephone calls (Cummins, Epps, Sethu, Breakspear, & Goecke, 2013; Mundt, Snyder, Cannizzaro, Chappie, & Geralts, 2007) or recordings of human interviews, reading or speech tasks. Although these tasks do not necessarily reflect the shorter human–VCA interaction, the review shows encouraging results in the use of vocal biomarkers for mental health assessments. 10.3.2.1.3 Biomarker for schizophrenia While Cummins and colleagues (Cummins et al., 2015) and Low and colleagues (Low et al., 2020) focused on the speech features, Corcoran and colleagues (Corcoran et al., 2018) pushed for linguistic analytic methods to predict schizophrenia. In particular, they observed psychotic speech to present lower but more varying semantic coherence (i.e., confusion in speech) and reduced presence of possessive pronouns. In fact, in schizophrenia, syntax tends to be less complex and the speaker can present speech flow derailment (Andreasen & Grove, 1986). It is, therefore, important to note the great potential in the implementation of speech analysis in VCAs for mental health. # Chapter 11: Design Considerations for Preparation, Optimization and Evaluation of digital therapeutics Page 135 - **Intervention design** -(how to design?) Asking how best to design treatments delivered through digital modalities intended to be used outside of traditional clinical environments. - **Study/evaluation design** -(how to answer the questions?) asking how best to answer key scientific questions about how the intervention could be optimised or whether it is achieving desired health impacts. **11.2 A framework for designing and evaluating digital therapeutic interventions 11.2.1 What do we mean by design?** [[What we are experiencing now, the going back and forth, non-linear progression, is part of the design process]] **Design is about balancing the desired outcome against the constraints, thus optimization is part of a good design.** - "In the context of attaining objectives under constraints, then, optimization is an integral feature of good design. In the intervention world, we generally think of optimization as maximizing effectiveness and societal benefit while working within resource constraints (Collins, 2018a, 2018b). Optimization studies are, thus, those that use data-driven approaches to evaluate different alternatives to determine which is optimal for a particular context, population, and set of resource limitations (Collins, Murphy, & Strecher, 2007, 2011; Riley & Rivera, 2014)." **11.2.2 The importance of optimization for digital therapeutic design** What guides optimization? - Questions that guide such optimization include (Bidargaddi, Schrader, Klasnja, Licinio, & Murphy, 2020; Liao, 2016; Nahum-Shani, Hekler, & Spruijt-Metz, 2015, 2018; Spruijt-Metz & Nilsen, 2014): Think about - What to deliver? - When and How to deliver. --- this author wrote about what should be done, but didn't explain or teach HOW to do it!! **11.2.3 A Conceptual framework for informing design decisions** - Make decisions based on the level of impact page 138![[Screenshot 2023-08-22 at 5.54.05 PM.png]] Page 139 -- The formative question- what problem are we trying to solve... ==we said we want to help patient in their recovery== may end up be different.. - Regardless of the methodological approach, the formative work stage helps researchers iteratively formulate and clarify the problem they are trying to solve. For our example intervention, researchers may come to the project with a broad understanding that, say, they are trying to improve functioning for persons with bipolar disorder. The formative stage, however, may bring about the realization that from the perspective of persons with bipolar disorder, the key problems are (1) consistent and timely access to providers, and (2) comorbid physical health conditions (Blixen, Perzynski, Bukach, Howland, & Sajatovic, 2016). As such, a project that initially may have focused on educating persons with bipolar disorder about their condition may be transformed through formative research into one where the central focus is on facilitating better access or communication with mental and physical health providers. The exact problem one should be trying to solve is rarely clear at the outset; the understanding of the problem changes and grows as researchers engage with the literature and the stakeholders. Formative research is the fundamental tool through which this process unfolds. page 141 talks about how to design what intervention to include, using factorial experiments **Use of MicroRandomised Trials (MRT) to test out different designs. Page 142-143** ![[Screenshot 2023-08-22 at 6.21.00 PM.png]]