Suicide Prevention Model Design Review 1

[TomRust]

Identity questions:

1. What care setting do you work in (PCMHI, GMH, SMH)
2. What is the primary service your team provides? (CC, MM, PSY, Group, Tele, Other)

Model validation questions: SP:

1. What barriers do you face in the effective management of high risk patients, i.e., getting patients in the “right setting, right treatment, right time”?

Stepped Care:

2. How do you manage patient care across care settings, e.g., with primary care, general mental health or specialty mental health?
3. Where do Vets “fall through the cracks”?
4. What are the barriers to transfer, after you decide to step a patient up?
5. What are the barriers to transfer, after you decide to step a patient down?
6. How can we get more patients to the right level of care?
7. Which new patients get priority for starting in your team - referrals from PCMHI or GMH or SMH?
8. What concerns do you have about implementing Stepped Care?

MBC:

9. What delays you deciding that a patient is ready for transfer?
10. How can we know what the “right level of care/right treatment” is for a Veteran sooner?
11. What concerns do you have about implementing Measurement-Based Care?
12. How will implementing Measurement-Based Care improve patient care and reduce patient risk?

Model use goals: • Get more patients in the right level of care • Reduce the number of patients who “fall through the cracks” i.e., who are waiting to start the right level of care? • Understand constraints to getting patients into the least restrictive care setting • Balancing workload across care settings, i.e., the whole continuum of care

Questions the model can address: (for ranking?)

1. What changes to clinical practice (e.g., how long to keep the average patient before referring them out?) will have the biggest impact on patient risk?
2. What changes to team operations (e.g., how to prioritize incoming referrals? how much to offer patients Choice?) will have the biggest impact on patient risk?
3. How will team practice changes effect the entire continuum of care (e.g, “Every patient PC takes is one more opening in GMH for one of their patients suffering an acute MH need”)?
4. How can we improve care across the entire continuum, instead of optimizing only one setting?
5. How to balance the trade-offs between access and quality?
6. Where have patients “fallen through the cracks” (i.e., are waiting to start the right level of care), and how can they best be addressed?
7. What are the system-wide impacts of improved collaboration between any two care settings?

[jamesmrollins]

Design Features Summary for SP UI By deadline, I will have graphical concept drawings that support the following design features.

1. Develop a black box "Pop-up" that will essentially be a popup box with widgets (stocks and flow). The use of the black-box popup will help simplify the model.
2. Provide feedback loop builds, as in previous models.
3. Provide General Mental Health, Specialty Mental Health and Primary Care modes for rates. These WILL NOT be displayed at the same time, rather one-at-a-time using a radio button.
4. Use a "Map Legend" feature to call-out certain sub-functions or over-arching functions that will also aid the simplification of the diagram.
5. Consider modification to Experiments tile: As read - Supply, Engagement Pattern and Sensitivity. Consider: Clinical Response, System Response and Sensitivity. Design Needs:
6. Feedback loop depths. How many depths do we want? Which loops and sliders do we want to display?
7. Team Data Table. I need a team data spreadsheet. What goes in the table?
8. Master Crosswalk Table with "i" information. We need a comprehensive mapping of model variables to Sim UI variables.
9. Do we want different colors for the loops?
10. Variable naming. Are variable names in objective form? Perception of Other's Needs (please confirm, modify, add or recommend deletion)
11. (mtl_facilitate_workgroup) Graphical files of new design features that can be used in lesson materials.
12. General support to the team for any explanations necessary to clarify design concept.

[lzim]

Thanks @TomRust and @jamesmrollins!

@TomRust This is a great list of questions. Fortunately, I think they have all been asked before. Therefore, the questions are too open-ended for our upcoming meeting. The SP module is also a culmination of over 3 years of stakeholder engagement. As a result, we have good answers to many of the questions already represented in our data UI/SPParams and previous model diagrams. This was evidenced by how similar our work to date is with other related efforts discussed on our call this morning.

@jamesmrollins Your question 7 concerns me a lot because it suggests a dangerous level rework already going on here. Tom and James, please review the SPParams tab of the ModelParameters.xlsx file for a reminder about language of existing standardized red variables, which were refined reviewed, along with additional definition details in column E: https://github.com/lzim/teampsd/tree/master/model_workgroup/models:

• Start Rates
• High Risk Flag Rates
• Time to Unflag
• Time from Flag to Referral
• Engagement Time Before Clean Transfer
• Engagement Time Before Quit
• Wait Times
• Symptom Proportions We need to tailor our questions to more efficiently resolve our remaining questions building about 1) key feedbacks, and 2) key experiments from these team red variables.

To simplify our SP module, and achieve a "saturation" regarding the key dynamics, I drafted revised questions based on the SPParams and the two files "CLD_Sept18.mdl" and "SuicidePrevention-18Sept2018.mdl" that help people to choose/prioritize team learning needs. We also need a much shorter overall list to go over in our stakeholder review meetings. We will triage based on areas of consensus and divergence just as we do in MTL session 4.

My summary points:

1. In addition to these questions and the model diagram, we will circulate the new data UI tabs/SPParams associated with the MTL 1.7 release SP module.
2. All questions should be specific to a) team data slider, or b) sensitivity test slider in the SP module diagram.
3. Keep in mind that the learners are still the multidisciplinary teams of providers that we've been preparing MTL to support from the beginning, between sessions 4 and 5 of the 12-session plan, they will choose this module over CC, MM, PSY and AGG, when there primary team learning goals and team needs at the intersection of their team with other teams.
4. Need to define they key terms we are using - ideally we would show where they are in the SP module prototype.
5. Following on our call this morning, all uses of PC as distinct from PC-MHI are intentional.
6. I included wait-times and referral acceptance, but I wasn't sure what to do with the "capacity" items from the .mdl files they are not in the SPParams tab - help @TomRust

I would love to develop response categories for the questions below before circulating that are operationally specific to our SP data, which has been standardized during the MTL 1.6 process. Can I get some help with that?

From the data UI SPParams variables and @TomRust "list of questions the model can address" - (LINDSEY PROPOSED EDITS): Overview: From the overall 6 domain VHA MH continuum of care - this is "zoomed in" on the interface between primary care (PC), primary care - mental health integration (PC-MHI), general mental health (GMH, e.g., BHIPs), and specialty mental health (e.g., PTSD Treatment Programs, SUD treatment programs and Telemental Health programs).

**Identity Questions:

1. In addition to team type, add a category for OMSHP leaders.
2. No need for a services question for this module as we assume perfect mixing of services in the SP module, and nearly all teams do a mix of services.
3. Add a discipline item: psychiatrist, psychologist, social worker, nurse, certified peer support specialist, adjunctive services (e.g., recreation/art therapy)

MBC Components: 1a. What information delays are associated with whether and when a team notices a patient is increasing in risk? 1b. What information delays are associated with whether and when a team notices a patient is improving/recovering? 2a. What factors influence the use MBC symptom information used in your team decisions and for coordinating care across teams (PC, PC-MHI, GMH, SMH) in your clinic? 2b. What delays/factors are associated with obtaining and using MBC? 3a. What team care decisions using standardized symptom measures will have the biggest impact on patient risk in your team's overall patient panel? 3b. What team care decisions using standardized symptom measures will have the biggest impact on patient recovery in your team's overall patient panel?

Median Engagement in Teams before Steps, based on Measures 4a. What information does your team have about how long your team keep the average patient engaged in your services when their symptoms aren't improving or are getting worse? 4b. What information does your team have about how long your team keep the average patient engaged in your services when their symptoms are improving and they are getting better? 5a. What information does your team have about how long your team keep the average patient before stepping them up? 5b. What information does your team have about how long your team keep the average patient before stepping them down?

Stepped Care Components: Stepping up 6a. What delays/factors influence team decisions about stepping a patient up to general mental health from primary care-mental health integration? 6b. What delays/factors influence team decisions about stepping a patient up to specialty mental health (e.g., PTSD Clinical Team, Additional Treatment Services) from general mental health? Stepping down 7a. What delays/factors influence team decisions about stepping down a patient to general mental health from specialty mental health (e.g., PTSD Clinical Team, Addiction Treatment Services)? 7b. What delays/factors influence team decisions about stepping down a patient to primary care from general mental health?

*Other Key Influences on Team Decisions Wait-times 8a. How does information about patients waiting to get into general mental health influence teams' decisions to step them up from primary care-mental health integration? 8b. How does information about patients waiting to get into specialty mental health influence teams' decision to step them up from general mental health? 9a. How does information about patients waiting to get into primary care influence teams' decisions to step them down from general mental health? 9b. How does information about patients waiting to get into general mental health influence teams' decisions to step them down from specialty mental health?

Referral Acceptance 10a. How do rejected referrals influence teams' decisions to step patients up from primary care-mental health integration to general mental health? 10b. How do rejected referrals influence teams' decisions to step patients up from general mental health to specialty mental health? 11a. How do rejected referrals influence teams' decisions to step them down from general mental health to primary care? 11b. How do rejected referrals influence teams' decisions too step them down from specialty mental health to general mental health?

*Symptom Proportions and Identified High-Risk Patients

12. What are the impacts of having a higher proportion of high symptom patients (among the overall team patient panel) on the quality of care in your team?

13a. What factors influence the use of high risk factor (HRF) information in your team? 13b. What factors influence the use of suicide screens and safety plans? 13c. What factors influence the use of Reach-Vet?

Thanks, let's keep this volley going to get this done by tomorrow :)

@TomRust @jamesmrollins For design, in GENERAL KEEP EVERYTHING AS ABSOLUTELY CONSISTENT AS YOU CAN WITH EVERYTHING WE'VE DONE SO FAR, I'd rule out the proposal for @jamesmrollins proposal 5 to change our categories. Note: We've already done a lot of iterations on all of this. See the Master Crosswalk 😅 RECOMMENDATION: Check early and often to avoid rework, check even earlier if you are deviating in any way, @staceypark and I can help us avoid rework.

THANKS!! 👍

Lindsey

[TomRust]

My thoughts on the "Design Needs" section:

6. How about we have 3 depths -- 1 with the stock and flow with the red data variables, 2 with the feedback loops, and 3 with the sensitivities and implement MBC and SteppedCare tests

7. Can we have the table only show one setting at a time, linked to the radio button selection?

8. Variable names (especially "New-ish Patients"!!!) definitely need a review

[lzim]

@TomRust and @jamesmrollins and @staceypark

Given the SPParams data table, response 7 and 8 are out-of-place. Please see my highlight about this above ASAP. 🙏 🤔 https://github.com/lzim/teampsd/tree/master/model_workgroup/models

6. We won't know the final # of intermediate depths for sure, until we know the key feedbacks or "stories." But, in general starting without feedbacks, and ending sensitivities will be consistent with our other MTL modules, which is what we should do: BE CONSISTENT.

7 and 8. I don't understand how we can present this SPParams data table one setting at a time, when some of the most critical definitions "Time from Flag to Referral" "Engagement Time Before Clean Transfer" and "Wait Times" are all about transitions from one setting to another. There is also no "newish patients" variable in SPParams file (only "Start Rate" is "newish patients" a Vensim Only -> sim UI variable not based on Team Data?)

NOTE: We will not have time to redo Andrew's prior work, so posts in this thread that deviate from SPParams variable names and definitions finalized before he left are not workable.

[jamesmrollins]

Hi @lzim @TomRust @staceypark @dlkibbe @branscombj @holbrooa @dlounsbu

Attached are preliminary drawings for Suicide Prevention for your consumption, digestion and feedback. These drawings will hopefully support the questions above as we craft our first stakeholder review. Remember that the PDF is a layered document. Please ensure you review the layers as we are conceptually adding a way to simplify the diagram, by allowing a "Drill Down" to other working parts of the model. Let me know if you have any questions.

Suicide_Prevention_v_1.1 copy.pdf

[jamesmrollins]

@TomRust @lzim

1. "Black Box" concept explained. The illustration below explains how the "Black Box" concept works to decrease the complexity of the drawing while maintaining the ability to drill down into more detail. Please note, that I still need to check with DEV to make sure it is possible to use widgets in popups.

2. Team Data table updated to reflect model parameters information contained in the SP tab.

3. My sense of "New-ish" patients are patients that are already in the system, but are transferred in. So technically, they are not "New" patients. "Patients New to Mental Health" is a bit long, so how about, "Patients Transferred In?"
   

4. Attached is an updated PDF with layers. Please ensure you click through the layers in order to see how the drill-down functions may work. (see highlights below)

5. @TomRust we need to also investigate how the UI might morph when the "SMH" or "PCMH" radio buttons are clicked. Further, we have some room where we could provide a legend that explains the interface on latch-on points between the three treatment delivery systems (i.e., PCMHI, GMH and SMH)

6. I am going to break away from this effort until I hear some feedback. In the meantime, I will work on the icon graphics files for issue #119 Suicide_Prevention_v_1.1 copy.pdf

[TomRust]

Since we're using the survey to narrow down our set of experiments learners can do with their team data, couldn't we design the survey directly around the team data table? Something along the lines of "How much does __ matter to getting patients in the right level of care?" And then have them put a "high, med, low" against each of the params?

[lzim]

Survey should be both to the diagram, which includes the red team data variables. Need to clarify key feedbacks and experiments is the goal. Looks like we're heading in the right direction. We are booked at 2PM, but can try to squeeze the start of our 3PM. Or, better just touch base at 4PM (all times Pacific) - sorry about the limited availability.

Lindsey

[TomRust]

SP Model User Survey - draft - 2018Sept26.docx

And I’d add an open ended question to each of these, asking respondents to list any other rates and times, or flows, or effects, that are important to (have a significant impact on) the ability to get and keep patients in the right level of care.

[jamesmrollins]

OK - here is the latest and the greatest. @TomRust I changed many of the variable names anticipating that we would need unique variables if we intend to make the widgets operate. Please take a close look and make sure I haven't over-stepped. We may even take a little time to go over all our variable names prior to survey release. Also, I didn't know what to do with "NEW?" Is that new patients? There was no rate associated with it (and I'm not looking for any :-) ) I've also attached an updated PDF with layers.

Suicide_Prevention_v_1.1 copy.pdf

[TomRust]

SP Model User Survey - draft - 2018Sept28.docx

Added question about impact of having PCMHI (along with implementing other initiatives)

[jamesmrollins]

@lzim @TomRust @staceypark @branscombj @holbrooa @dlkibbe Good Evening Everyone, Attached is the draft PDF of the work-shopped simulation design for 1.70. It is a PDF built with layers, so you have to turn on and off items in the appropriate sequence to get the effect. Good news is that i can get the minimized expanded outputs screen (text only) into the default view. We can also have the all the drill downs open at the same time. Experiment sliders will be manipulable while these popup windows are open.

Lindsey mentioned working again on this on Monday. I didn't however, see a meeting notice. I will try to reveille by 0800 hours, but I'm not making any promises!! Suicide_Prevention_v_1.2.pdf

Night! James

[lzim]

Thanks @jamesmrollins

@staceypark and @TomRust I can meet 10:30AM-11:30AM Pacific today if we need it. I've been working for a few hours and have yet to take a look at this update.

@branscombj and @dlkibbe Please disregard for now as you are focused on moving the EES master brochures and post-tests forward with the ees_workgroup.

Thanks!

[lzim]

@TomRust @jamesmrollins and @staceypark

I will review before our 7AM Pacific tomorrow. Stacey will schedule a Lucid. So far, it looks really good. Thanks for all the hard work! :sweat_smile:

Note: I am re-thinking our order for the stakeholder review.

I now think the most effective approach will be to schedule a meeting during which we go through the data and diagram and have people respond to the survey live during the demo/walk through.

We can discuss it tomorrow.

Thanks!

Lindsey

[jamesmrollins]

Oh! That is a great idea!! We should rehearse, though.

Get Outlook for iOShttps://aka.ms/o0ukef

---

From: Lindsey Zimmerman notifications@github.com Sent: Monday, October 1, 2018 3:07 PM To: lzim/teampsd Cc: James Rollins; Mention Subject: Re: [lzim/teampsd] Suicide Prevention Model Design Review 1 (#199)

@TomRusthttps://github.com/TomRust @jamesmrollinshttps://github.com/jamesmrollins and @staceyparkhttps://github.com/staceypark

I will review before our 7AM Pacific tomorrow. Stacey will schedule a Lucid. So far, it looks really good. Thanks for all the hard work! 😅

Note: I am re-thinking our order for the stakeholder review.

I now think the most effective approach will be to schedule a meeting during which we go through the data and diagram and have people respond to the survey live during the demo/walk through.

We can discuss it tomorrow.

Thanks!

Lindsey

— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHubhttps://github.com/lzim/teampsd/issues/199#issuecomment-426080415, or mute the threadhttps://github.com/notifications/unsubscribe-auth/AcvHMS_bQpGVSD_Bks08vvRtjfIwM1YQks5ugpIfgaJpZM4W3PfS.

[jamesmrollins]

Hi @TomRust - Here is a drawing of the sketch you provided. The yellow highlights indicate areas where I did not give portray the feature exactly as you sketched. I was concerned, because to "To PC" and "To SMH" rates are actually two step processes in the main diagram, where first a signal is given, then a patient move occurs. I have also posted a copy of the sketch for the official record.

Let me know if you have any comments or questions.

Thanks, James

[jamesmrollins]

@TomRust @lzim @staceypark here are the latest designs:

And the PDF. Lindsey, please take a look at the Team Data Table and see if it meets expectations. Suicide_Prevention_v_1.2.pdf

[lzim]

Thanks @jamesmrollins full day today, but a very good day overall. Data Table looks GREAT to me.

@TomRust and @staceypark - Even with the questions we discussed yesterday afternoon regarding the "Implement MBC" and "Implement Stepped Care" green Sliders.I think we're ready for stakeholder review.

We will work to organize the survey questions to following the design system stories. For a walk through meeting.

THANK YOU!

Lindsey

[jamesmrollins]

Hi @TomRust @lzim @staceypark , My goal is to get DEV kicked off tonight. I have enough to go with, but taking a few minutes to review and validate the pictures below would help save a little time and money in the long-run. Also, I need polarities on arrows. THANKS!!

1. This is the base view (I only did PC/PCMHI. Once I get information I can finish the rest). On what layer do we want to place the "PC/PCMHI Patient Load" variable (see yellow highlight).

2. Work Pressure Delays Care Decisions loop.

3. Higher Care Quality Improves Recovery loop.

4. Work Pressure Reduces Access loop.

5. Access Pressure Increases Patient Starts loop:

6. Burnout Increases Clinician Quits loop:

7 Sensitivity Sliders:

[jamesmrollins]

@lzim There is an updated PDF that contains the builds attached. I still need arrow polarities.

I released a design to DEV so we can get started. See below for our proposed schedule:

1. Complete Design Document (ver 1): NLT 8 October 2018. Note: The ver 1 document is an 85% design, so changes will occur. Changes will be tracked in the next version of the design document (see item 4 below).
2. Notice to Proceed to Development Team, NLT: 9 October 2018
3. Stakeholder group review: ~15 October 2018 (this is a week-long process, may be shorter)
4. Design Document (ver 2): 22 October 2018.
5. Module in Test: 09 November 2018.
6. Testing Complete ~ 16 November 2018

Suicide_Prevention_v_1.3.pdf

[TomRust]

@jamesmrollins, @lzim: I was putting polarity signs on all the arrows in the UI ... and realized that we have an important question to answer before I can complete that task:

Do we want the set of "Implement x" experiment variables to point directly at the rates they affect or point at the times controlling those rates?

Currently in Vensim, those experiment variables affect the times (e.g., a full Implementation of MBC will cut the "Time to Detect" in half, which will cause the rate to double). To me, this keeps the "physics" of the model clear -- you can see our causal theory about how implementing MBC will improve the Detection Rate. If the "Implement x" experiment variables point directly at the rates, then it implies that those rates can change without providers changing their "Time to Detect" -- implementing MBC must be changing something else that influences the Detection Rate that isn't shown in our model.

However, this makes for a more confusing UI diagram -- as green slider variables are now pointing at red slider variables. Also, we'll need more outputs, too, as (in the example above) "Time to Detect" is no longer solely an input, but is also an output, thus, we'll need to include it in the Expanded Outputs drop down list.

To me, that additional clarity is worth the "facilitator lift" of explaining sliders pointing at sliders. But, I'm bias. :)

What do you all think?

[dlkibbe]

@TomRust just curious in your opinion which option would make it easiest for the non-systems model expert to understand?

[jamesmrollins]

@TomRust @lzim @dlkibbe

Pursuant to Debbie's observation, it seems that consistency would be helpful. The capture below is the only instance where an "Implement X" variable is pointed to a team data slider variable. I believe that the "Time to Recommend Step up" variable should be a direct control as in the other cases, and the "Implement X" slider should be pointing directly at the rate. So this:

Should look more like this:

There are two instances where the Implement MBC slider points at a quality variable. My understanding is that Implement MBC is a time conserving function. But, I am not sure what the team's definition of quality is here. In industrial settings, quality is typically a measurement of the degree of conformance to standards. Time and cost are two separate measurements. Thus, I want to conclude that it is not relevant to point Implement MBC to quality. In my experience, speeding things up often has a negative effect on quality.

From the main UI:

From the Effect of Measurement Based Stepped Care on Patient Symptoms and Risk drill-down:

Thanks, James

[TomRust]

@jamesmrollins -- I totally agree that the diagram should be consistent -- either all point to the rate directly, or all point to the variable controlling the relevant rate.
@dlkibbe -- It's a question of how much detail do we want in the diagram: If we point the "Implement x" variables directly at the rates, then the diagram is simple. But, the mechanisms involved (ie, that implementation is changing providers' average "Time to Detect," or their patients' average "Time to Improve," etc.) will still need to be explained. To me, putting all of that explanation on the facilitators makes it more difficult for the users to understand how the implementation of either MBC or Stepped Care actually change any of the rates. I'd want the visual to match the verbal explanation.

[TomRust]

@jamesmrollins -- Whatever we decide, there are still a few general rules for arrow polarities: 1) all arrows pointing AT an "effect" variable should be "+" 2) all arrows pointing FROM a "desired time" variable should be "-" 3) all arrows pointing FROM a "time" variable should be "-" (but this is trumped by rule 1) 4) all arrows pointing FROM a "sensitivity" variable should be unmarked

[TomRust]

@jamesmrollins -- here is a pdf of GMH with the polarities. The loops in the other services should be the same.
Scan.pdf

[jamesmrollins]

@TomRust Thanks Tom, I will update the "official" design document.

[jamesmrollins]

@lzim @TomRust Below are some design components where I have given preliminary guidance to the DEV team:

1. Design feature: Home screen function where the user selects what team they are on (i.e., GMH, SMH or PC/PCMHI) in order to direct their focus to their own model area.

Guidance: This is problematic since the model presents all three team areas simultaneously. Therefore, their isn't any existing simulation architecture to support limit or otherwise manage the focus of the team. This may be technically feasible, but will require a significant design effort (therefore more time). Therefore, we should push to 1.8 or drop altogether (bang for the buck).

2. Design Feature: Export files

Guidance: Export files will contain data for all model areas (no matter who is asking) that are separated by a break-line and title. This will require no additional coding; therefore save time.

3. Design Feature: Charts available in the Expanded Outputs Screen.

Guidance: Provide charts for selected mental health area (GMH, SMH or PC/PCMHI) only. For example, when GMH is selected, charts for GMH will be available. For between MH area analysis, these charts will be available in the MH Care Continuum mode.

[TomRust]

@lzim @jamesmrollins -- these all make sense to me.

[TomRust]

@jamesmrollins -- Issue with experiment slider "% opening to ..."

This slider is actually 3 sliders -- it controls the % of openings allocated to the three possible sources of new patients. It needs to look like what we have for "Starters who Initiate %" in PSY, so people don't accidentally create more supply by these three percents summing to more than 100%.

We'll also need to put the BC values into the Team Data Table...if they aren't already there.

[TomRust]

@jamesmrollins -- Do we have an inflow straight into the stock of "GMH Patients Early in Care" from InPt?

We should! Its a green rate (like "Community Care Rate") and needs an experiment slider.

This only applies to GMH.

[jamesmrollins]

@TomRust @lzim @staceypark @dlkibbe @branscombj @dlounsbu Hello Everyone! Here is a flat-file of the latest Suicide Prevention Model. This enables side-by-side comparison and in many ways is easier to navigate. I am trying to also publish a stacked file, but my computer is screwing up the layers. For some reason it collapsed everything into one layer. This will take some time to get resorted. When I get it sorted, I will publish it here.

Suicide_Prevention_v_1.4 - Flat.pdf

[jamesmrollins]

@TomRust @lzim @staceypark @dlkibbe @branscombj @dlounsbu OK - here is the rebuilt PDF that has working layers to support a narrative. I optimized it for web, so hopefully loads and runs ok. Tested out ok for me. Let me know if there are any problems.

Thanks, James suicide_prevention_v_1.4_layered.pdf

[staceypark]

THANK YOU @jamesmrollins !! :smile: :heart: I'm working on the survey right now

[lzim]

Hi @TomRust! In advance of our call tomorrow have you made any of the edits to the SPParams tab that we discussed on Wed? Just checking :)

Thanks! https://github.com/lzim/teampsd/blob/master/model_workgroup/models/ModelParameters.xlsx

[jamesmrollins]

Team - please confirm that we only have one basis of comparison view in the Expanded Outputs. The experimenter should only be able to compare previous runs to the current run, correct. This is an important question, if we need to compare say, GMH results with SMH results in any of the GMH, SMH or PC/PCMHI comparison modes.

[jamesmrollins]

@TomRust @lzim

Ok - here is the GMH sample file. A flat picture is below and a layered PDF is attached. Give me feedback please, as I will resume work at 1200 tomorrow.

Thank you, James

Suicide_Prevention_v_1.5 demo.pdf

[TomRust]

Looks great!!!! The new default looks like it was meant to be. I love the stacked cards in the GMH view for "Openings for new patients %."

Two tiny changes to the Model Diagram Section: Reverse the polarity of the arrows coming from "GMH High Symptom %" Add negative signs to the arrows coming from "GMH Time to Detect"

Five small changes to the Experiment Section: Change the units on the "GMH Patient Panel" slider from Appts to Pts Add a "BC" button to the "GMH Patient Panel" slider Add a second green slider for "GMH Community Care Rate" -- we need one for high and one for low symptom pts Add a slider for "GMH Time to Starting Community Care" to the section Move the slider for "GMH Implement MBC" to the top of the set

[jamesmrollins]

@TomRust @lzim Thanks Tom. I will get to work on the rest of them. I also you to take a look at the question I asked about the expanded outputs section. I am assuming we only need a compare experiments mode for a single service, but not across services, correct?

[lzim]

Thanks @TomRust will have my final notes posted for you here shortly @jamesmrollins

[TomRust]

Diagram change in GMH Settings Base View: "GMH and PC/PCMHI Implement Stepped Care" should point at "GMH to PC/PCMHI Time to Recommend Step down"

Polarity signs for the GMH Setting Base View: Arrow from "GMH Time to Inpatient or Residential" = "-" Arrow from "GMH Time to Ending" = "-" Arrow from "GMH Implement MBC" = "+" Arrow from "GMH and PC/PCMHI Implement Stepped Care" = "-" Arrow from "GMH and SMH Implement Stepped Care" = "-"

[TomRust]

Sorry @jamesmrollins ! I didn't see that question earlier. NO, I don't think we'll need the feature that allows users to compare across settings in the same chart.

[lzim]

Thanks again @jamesmrollins - I'm actually going to post a few comments piece by piece, but first... 😃

For your question about the Control Panel. "Services" are mixed in this module, do you mean "Settings?" (i.e., GMH, SMH, PC/PCMHI). For efficiency, will they need a way of selecting between these, just as they need to select their "Setting View." I know we named variables in a way that we have to take into account when we think about this (e.g., GMH Low Symptom Patients, SMH Low Symptom Patients.

Besides comparing runs, there is likely a useful selection that makes the most sense in this module. For example, we currently have:

• MM - compare diagnostic cohorts
• PSY - expand the Sankey
• AGG - compare services These all fit the purposes/learning objectives of these 3 models very well. Having these interactive options helps to make evident the high-yield decisions or the key insights of each model.

I wonder if the analogous thing would be to compare the key transitions in this module. For example, from the GMH point of view:

• Compare stepping up variables and stepping down variables to see how these tradeoff easily
• Putting the same concept another way: Compare GMH -> PC/PCMHI (stepping down) and GMH -> to SMH (stepping up)
• EDIT: Adding a third related concept: Compare "high symptom patients" "high risk patients" and "low symptom patients."

@TomRust given your work with the model, are there comparisons of key rates or stocks that you expect would really help learners put the full picture together?

[jamesmrollins]

@TomRust @lzim Yes, I do mean settings. The system is currently set up to follow the settings selection in the Experiment Timeline section. So, if the user selects GMH, then they would get all the GMH-prefix charts. I recommend you handle transitions in the Mental Health Continuum setting.

[TomRust]

@jamesmrollins -- one more catch in the GMH Setting Base View:

"GMH Effect of PC/PCMHI Wait Time on Step down Rate" should have the word "Recommend" in it

[lzim]

Sorry for the delays:

@jamesmrollins I really like the new default view!! "GMH Effect of Measurement Based Stepped Care on Patient Symptoms and Risk" A minor edit, and a couple diagram checks with Tom, then I have comments about Experiments for you below.

• Need to edit red sliders in model diagram current labeled "GMH Flag Rate %" and "GMH Time to Unflag" to use the full descriptors in diagram and Experiments (as you have for one of them): "GMH High Risk Flag Rate %"

@TomRust Questions about this diagram: is this polarity path correct in this view?

• GMH Implement MBC -> GMH Care quality (+) - makes sense - more MBC, better quality (or vice versa)

• GMH Care Quality -> GMH Time to Unflag (+) - confuses me? Wouldn't this be opposite (-): higher care quality reduces time to unflag, and lower care quality increases time to unflag?

• GMH Time to Unflag -> GMH High Risk Unflag rate (-) - makes sense - more time to unflag, slower unflag rate (or vice versa)

• Arrows between "GMH Time to Detect" and the two rates, shouldn't these have polarities?

• I'm not sure I'm getting the opposite polarities from the proportion "GMH High Symptom %" on the two rates. Seems like, because high symptom patients are likely to take more time to mange, the higher the proportion of High Symptom patients the slower the rate of detection for both High and Low symptom patients.

@jamesmrollins I see in Experiments you moved the "Effect of Measurement Based Stepped Care on Patient Symptoms and Risk" to the top, which is great.

• Within the two large sub-section of Experiments (i.e., "within" meaning keep the two sub-sections: "Effect of Measurement Based Stepped Care on Patient Symptoms and Risk" and "General Mental Health," I would really like to group green sliders together, and red sliders together. I know you are tracing the model diagram, but I don't think that aids clarity. Rather going back and forth between green and red makes it more complex visually, which increases cognitive demand.

• Need to update the sliders for Community Care to reflect the updated model diagram: Add:

• GMH High Symptom Patients Community Care Rate

• GMH Low Symptom Patients Community Care Rate

• GMH Delay to Starting Community Care Rate

Sensitivity

• Put "GMH Sensitivity of Care Quality to Patient Load" as the first/upper left slider, as it is the only viewable from our new default view.
• Consider whether it may also be helpful to have it be the only slider in the top row of the sensitivities, then have below that 3 rows of purple sensitivities for all the sliders that are in the GMH Setting View.

Okay, I'll make a separate post about the GMH Setting View in just a sec :eyes:

Thanks!

[lzim]

GMH Setting Base View

• Need to add two the default view, the arrow from "GMH Effect of Measurement Based Care on Patient Symptoms and Risk" back up to "GMH Patient Load" and make it navy blue to close the loop "Higher Care Quality Improves Recovery."

• This is the key reinforcing feedback loop story "Higher Care Quality Improves Recovery" told in the new default view: "GMH Effect of Measurement Based Care on Patient Symptoms and Risk."

• And it should be in default view, since you've removed it from the "Reveal Complexity" in the Experiment Timeline"

• NOTE: Delete "Section" from the Experiment Timeline - Just keep it "Experiment Timeline"

• Recommend that we center top to bottom the text for "GMH Effect of Measurement Based Care on Patient Symptoms and Risk."

Experiment Sliders

• Right now, the green slider "GMH Implement MBC" is duplicated because it occurs in the "Effect of Measurement Based Stepped Care on Patient Symptoms and Risk" and "General Mental Health" sub-sections.

Proposed options A and B: a. Add a sub-section header to Experiments entitled, "Measurement Based Care" and include the green slider "GMH Implement MBC" there for use in either default or setting view. b. Keep the duplicate, but make the first green slider under "General Mental Health"

• Need to add green slider "GMH Step Down from Inpatient Rate"
• When green sliders are grouped together, please put side by side the two "Implement Stepped Care" sliders.

Okay, that's all I see - GREAT Work 👏

[jamesmrollins]

@lzim Lindsey, regarding your comment, "Need to update the sliders for Community Care to reflect the updated model diagram." My copy shows that as completed. Do you have the latest file?

James

[jamesmrollins]

@lzim @TomRust Regarding "Implement MBC" slider. Although I show two sliders, they are logically the same measurement (so we only need one). If MBC can be different between the two places it affects in the diagram, then we will need unique names for them. The red slider "GMH Time to Ending" is also duplicated in two places. But there is only one variable to adjust.

[jamesmrollins]

and the closed loop. . .