0) Changelog 6/29

-reversed polarity of all “Mental health” edges and then recoded to “Mental health disorders” -recoded “Healthy diet” and “Quality diet” to “Access to quality diet” -recoded “Level of education” to “Education level” -recoded “Quality job” to “Access to quality jobs”

There are now 362 nodes, please see section 3) Union model for details.

It was hard for me to find “major” action ideas because there was less agreement on these. One way to think about it might be: what are the most common parts of the core model that action ideas work upon?

ARCHES

0) Goals:

Characterize a unified structure that all groups were trying to describe (because stratification by age, gender, and deprivation status was difficult with this small number of groups)

1) Group structure

Demographic data for these groups is mostly overlapping.

-Note: group numbering used here is sequential, so may not match the numbering used in the folder structure – be sure to check against model name).

-Note: figures show distribution “violins” and also points for each person. Points are shown with a small amount of x and y axis noise (jitter) just to be visible / not overplotted.

Group n_people birth_year gender deprivation model
1 8 1963.75 1.75 0.12 ./CLD-17-03-23-Standardized-v2_Ram_Updated.mdl
2 11 1962.27 1.91 0.27 ./CLD-18-3-23_LoopsColored_ActionIdeas-JFT_RAM_Updated.mdl
3 6 1953.17 1.83 0.33 ./CLD.3.19.2023_MM-AW_RAM_Updated.mdl
4 6 1954.17 2.00 0.00 ./CLD-19-03-23-G2_JF_RAM_Updated.mdl
5 8 1955.88 1.88 0.25 ./ARCHES GMB Vensim Number 01 03.24.2023_Written by WZ SL_JF_RAM_Updated.mdl
6 7 1953.43 1.86 0.57 ./ARCHES GMB Vensim Number 02 03.24.2023_Written by CC_Meena&Alexis_Ram_Updated.mdl
7 7 1953.14 1.86 0.43 ./ARCHES GMB Vensim Number 01 03.25.2023_Written by WZ SL RAM_Updated.mdl
8 6 1962.17 2.00 0.17 ./GMB-JFT-NR-03-25-23-v2_Ram_Updated.mdl

2) Models

Each group proposed a model.

-In example plots, light blue indicates “increase” and light pink “decrease”. Dark colors are Action Ideas.

Group number_edges number_unique_nodes
1 88 48
2 117 86
3 100 82
4 115 75
5 127 69
6 132 84
7 108 61
8 154 81

3) Union model

When all models are overlayed, we can build a “union” model where every edge is represented. This comprises 841 unique combinations of cause, effect, and polarity. These includes 469 ‘increase’ edges and 270 ‘decrease’ edges (not including Action Ideas) as well as 91 ‘increase’ edges connected to Action Ideas, and 11 ‘decrease’ edges connected to Action Ideas. Edges connect in various ways 362 nodes (vertices).

This is a big, complex network that is difficult to display or really contextualize.

Only a small number of edges are reported by multiple groups, suggesting that groups are describing different areas of the system: 2 edges were mentioned by 6 groups, 3, by 5 groups, 3 by 4 groups, 13 by 3 groups, 43 by 2 groups, and the remaining 777 by only one group.

Here’s the model with just edges mentioned by at least two groups:

4) Core model

This union model may be further simplified by A) considering only loops/cycles (so no ‘externals’ or ‘dead end’ nodes) and B) only considering those that connect to the central node/vertex of concern, in this case Dementia Risk. This gives us finally 18 nodes/vertices, connected by 39 edges/vectors.

I think this is a nice “Core Model” for our descriptive paper.

Here’s the edgelist for this core model

cause pol effect number_models models
Stress increases Mental health disorders 6 3,4,5,6,7,8
Access to quality diet decreases Chronic disease 5 4,5,6,7,8
Chronic disease increases Dementia risk 5 1,3,4,5,6
Access to quality jobs increases Income 4 1,2,3,4
Exercise decreases Chronic disease 4 1,4,7,8
Dementia risk increases Mental health disorders 3 1,5,8
Exercise increases Physical health 3 2,3,7
Family caregiving decreases Access to quality jobs 3 1,4,5
Food desert decreases Access to quality diet 3 1,5,6
Health literacy increases Access to quality diet 3 1,4,5
Health literacy increases Exercise 3 1,4,5
Health literacy decreases Stress 3 6,7,8
Mental health disorders increases Dementia risk 3 1,5,8
Stress increases Chronic disease 3 5,6,8
Stress increases Substance abuse 3 5,7,8
Substance abuse increases Mental health disorders 3 3,5,6
Access to quality diet increases Physical health 2 7,8
Access to quality healthcare decreases Dementia risk 2 3,4
Crime increases Food desert 2 4,8
Crime decreases Safe neighborhood 2 6,8
Crime increases Stress 2 6,7
Dementia risk increases Family caregiving 2 2,8
Dementia risk increases Health literacy 2 1,2
Dementia risk increases Social isolation 2 1,8
Family caregiving increases Health literacy 2 1,8
Family caregiving decreases Income 2 2,4
Family caregiving increases Stress 2 5,8
Food desert increases Access to quality diet 2 4,8
Health literacy increases Family caregiving 2 1,4
Income increases Access to quality diet 2 6,8
Income decreases Crime 2 6,8
Income increases Private transportation 2 1,2
Mental health disorders increases Crime 2 5,8
Physical health increases Family caregiving 2 1,7
Private transportation increases Access to quality healthcare 2 1,7
Safe neighborhood increases Exercise 2 4,8
Social isolation increases Mental health disorders 2 1,5
Stress increases Dementia risk 2 4,6
Substance abuse increases Crime 2 5,8

Running an ordination, there’s still no significant difference by demographics among groups by which core model edges they report, but here’s a comparative chart and hierarchical clustering showing how similar groups are to one another based on the core model edges they report.

4.2) Core model - intervening nodes

Do other models show these same edges, but with intervening nodes?

Below I show the same table, but with the indirect paths taken by additional models to join the same cause and effect. Column “models with intervening” shows groups that reported the same edges, but with intervening nodes. Column “intervening” shows the group number, the overall polarity (+ or -) and then the shortest path from cause to effect.

NOTE two warnings here:

1: the path reported is just the shortest path from cause to effect for the group, there may be and often are many other longer paths, some could have different polarity.

2: in some cases where the polarity is different it seems to be to be a mistake, e.g.in “6(-): Crime - Safe neighborhood + Food desert”, it seems counterintuitive that “Safe neighborhood” would increase “Food desert”, or in “6(+): Family caregiving + Early retirement + Stress + Lack of interest in health knowledge + Health literacy” it seems counterintuitive that “Lack of interest in helath knowledge” would increase “Health literacy”. These last errors (if they are) should probably be corrected but are few at this point and I don’t believe they change the overall outcome

cause pol effect number_models models models_with_intervening intervening
Stress increases Mental health disorders 6 3,4,5,6,7,8
Access to quality diet decreases Chronic disease 5 4,5,6,7,8 2, 3 2(-): Access to quality diet + Good lifestyle - Genetics + Dementia risk + Family caregiving - Income + Private transportation + Regular health visits - Chronic disease, 3(-): Access to quality diet - Obesity + Chronic disease
Chronic disease increases Dementia risk 5 1,3,4,5,6 2, 7, 8 2(+): Chronic disease + Side effects of medication + Dementia risk, 7(+): Chronic disease - Physical health + Family caregiving + Private transportation + Education level + Socialization - Dementia risk, 8(+): Chronic disease - Brain health - Dementia risk
Access to quality jobs increases Income 4 1,2,3,4
Exercise decreases Chronic disease 4 1,4,7,8 2, 5 2(-): Exercise - Mental health stigma + Dementia risk + Family caregiving - Income + Private transportation + Regular health visits - Chronic disease, 5(-): Exercise - Dementia risk + Mental health disorders - Access to quality diet - Chronic disease
Dementia risk increases Mental health disorders 3 1,5,8 4 4(-): Dementia risk + Out of pocket cost of expenses + Poverty + Intergenerational living - Mental health disorders
Exercise increases Physical health 3 2,3,7 1, 8 1(+): Exercise - Chronic disease + Family caregiving - Physical health, 8(-): Exercise - Chronic disease - Brain health - Dementia risk + Family caregiving + Health literacy + Physical health
Family caregiving decreases Access to quality jobs 3 1,4,5 2, 7, 8 2(+): Family caregiving - Income + Private transportation + Regular health visits - Chronic disease + Side effects of medication + Dementia risk + Out of pocket cost of expenses + Poverty + General political apathy - Political influence + Racist state policy + Biased criminal justice system + Biased people in criminal justice system - Access to quality jobs, 7(+): Family caregiving + Private transportation + Education level + Access to quality jobs, 8(+): Family caregiving + Stress - Job productivity - Access to quality jobs
Food desert decreases Access to quality diet 3 1,5,6 7 7(-): Food desert - Access to quality food + Access to quality diet
Health literacy increases Access to quality diet 3 1,4,5 7, 8 7(+): Health literacy - Stress - Physical health + Family caregiving + Private transportation + Education level + Access to quality jobs + Economic security + Access to quality food + Access to quality diet, 8(-): Health literacy - AI Junk food consumption + Access to quality diet
Health literacy increases Exercise 3 1,4,5 8 8(+): Health literacy - Substance abuse + Crime - Safe neighborhood + Exercise
Health literacy decreases Stress 3 6,7,8 5 5(-): Health literacy + Exercise - Dementia risk + Mental health patient dangerous behavior + Family caregiving + Stress
Mental health disorders increases Dementia risk 3 1,5,8 3, 7 3(+): Mental health disorders - Brain activity - Dementia risk, 7(+): Mental health disorders - Family caregiving + Private transportation + Education level + Socialization - Dementia risk
Stress increases Chronic disease 3 5,6,8 4, 7 4(+): Stress + Dementia risk + Out of pocket cost of expenses + Caregiver stress + Chronic disease, 7(+): Stress - Physical health + Family caregiving + Private transportation + Education level + Access to quality jobs + Economic security + Access to quality food + Access to quality diet - Chronic disease
Stress increases Substance abuse 3 5,7,8 3 3(-): Stress + Mental health disorders - Substance abuse
Substance abuse increases Mental health disorders 3 3,5,6 8 8(-): Substance abuse + Crime + Food desert + Access to quality diet + Physical health - Mental health disorders
Access to quality diet increases Physical health 2 7,8 1, 3 1(-): Access to quality diet + Chronic disease + Family caregiving - Physical health, 3(+): Access to quality diet - Obesity - Physical health
Access to quality healthcare decreases Dementia risk 2 3,4 1, 6, 8 1(-): Access to quality healthcare + Access to healthcare information + Health literacy + Exercise - Chronic disease + Dementia risk, 6(-): Access to quality healthcare - Misdiagnosis + Stress + Dementia risk, 8(-): Access to quality healthcare - Chronic disease - Brain health - Dementia risk
Crime increases Food desert 2 4,8 5, 6 5(+): Crime + Benefits + Food desert, 6(-): Crime - Safe neighborhood + Food desert
Crime decreases Safe neighborhood 2 6,8
Crime increases Stress 2 6,7 5, 8 5(+): Crime + Prison population - Parental guidance - Foster care system + Stress, 8(+): Crime - AI WU mobile clinics + Access to quality healthcare + Health literacy - Stress
Dementia risk increases Family caregiving 2 2,8 1, 5 1(+): Dementia risk + Health literacy + Family caregiving, 5(+): Dementia risk + Mental health patient dangerous behavior + Family caregiving
Dementia risk increases Health literacy 2 1,2 8 8(+): Dementia risk + Family caregiving + Health literacy
Dementia risk increases Social isolation 2 1,8 2 2(-): Dementia risk + Out of pocket cost of expenses + Poverty + General political apathy - Political influence + Racist state policy - Quality housing - Family structural breakdown + Social isolation
Family caregiving increases Health literacy 2 1,8 2, 6, 7 2(+): Family caregiving - Income + Private transportation + Regular health visits - Chronic disease + Side effects of medication + Dementia risk + Health literacy, 6(+): Family caregiving + Early retirement + Stress + Lack of interest in health knowledge + Health literacy, 7(-): Family caregiving - Caregiver leisure time - Research + Health care professional miscommunication - Equity of delivery in healthcare - Lack of reliable information - Health literacy
Family caregiving decreases Income 2 2,4 1, 8 1(-): Family caregiving - Access to quality jobs + Income, 8(-): Family caregiving + Part time job - Income
Family caregiving increases Stress 2 5,8 6, 7 6(+): Family caregiving + Early retirement + Stress, 7(+): Family caregiving - Caregiver leisure time - Research + Health care professional miscommunication - Equity of delivery in healthcare - Lack of reliable information - Health literacy - Stress
Food desert increases Access to quality diet 2 4,8 7 7(-): Food desert - Access to quality food + Access to quality diet
Health literacy increases Family caregiving 2 1,4 2, 5, 7, 8 2(-): Health literacy - Mental health stigma + Dementia risk + Family caregiving, 5(-): Health literacy + Exercise - Dementia risk + Mental health patient dangerous behavior + Family caregiving, 7(+): Health literacy - Stress - Physical health + Family caregiving, 8(-): Health literacy - Stress + Mental health disorders + Dementia risk + Family caregiving
Income increases Access to quality diet 2 6,8 1, 4 1(+): Income - Social isolation - Access to quality diet, 4(-): Income - Poverty + Food desert + Access to quality diet
Income decreases Crime 2 6,8 2 2(+): Income + Private transportation + Regular health visits - Chronic disease + Side effects of medication + Dementia risk + Out of pocket cost of expenses + Poverty + General political apathy - Political influence + Racist state policy + Biased criminal justice system + Biased people in criminal justice system - Access to quality jobs - Crime
Income increases Private transportation 2 1,2 4 4(+): Income - Poverty - Private transportation
Mental health disorders increases Crime 2 5,8
Physical health increases Family caregiving 2 1,7 8 8(-): Physical health - Mental health disorders + Dementia risk + Family caregiving
Private transportation increases Access to quality healthcare 2 1,7
Safe neighborhood increases Exercise 2 4,8
Social isolation increases Mental health disorders 2 1,5 3, 6, 8 3(+): Social isolation - Community network - Mental health disorders, 6(+): Social isolation + Depression + Mental health disorders, 8(+): Social isolation + Stress + Mental health disorders
Stress increases Dementia risk 2 4,6 3, 5, 7, 8 3(+): Stress + Mental health disorders - Brain activity - Dementia risk, 5(+): Stress - Access to quality diet - Dementia risk, 7(+): Stress - Physical health + Family caregiving + Private transportation + Education level + Socialization - Dementia risk, 8(+): Stress + Mental health disorders + Dementia risk
Substance abuse increases Crime 2 5,8 2 2(-): Substance abuse - Good lifestyle - Genetics + Dementia risk + Out of pocket cost of expenses + Poverty + General political apathy - Political influence + Racist state policy + Biased criminal justice system + Biased people in criminal justice system - Access to quality jobs - Crime

4.3) Core model - action ideas

How did action ideas proposed by each group act on elements of the core model?

group core nodes targeted by AIs
1 Chronic disease, Family caregiving, Social isolation, Access to quality healthcare, Health literacy
2 Health literacy, Access to quality diet
3 Dementia risk, Health literacy
4 Crime, Health literacy, Access to quality healthcare
5 Substance abuse, Access to quality diet, Exercise, Chronic disease, Health literacy
6 Family caregiving, Stress, Substance abuse
7 Stress, Family caregiving, Dementia risk, Access to quality diet, Exercise, Health literacy
8 Access to quality diet, Health literacy, Chronic disease, Income, Substance abuse, Access to quality healthcare