Summary of results

Model inputs

Three main modelling methods were attempted (generalised linear modelling (GLM), using a quasibinomial model with a logit link function; random forest (RF) model modelling the outcome; and a random forest model that models the change in the outcome from the previous year).

Variations on model inputs were also tested. Models were tested where the input data varied between 3 and 7 years. For each of those, different amount of lagging to the input data was tested too - in order to see whether there was a knock on effect of a previous year for an input variable on the outcome variable for the following year. Models with 0, 1 and 2 lagged years were tested. The importance of the previous year’s outcome value as a predictor variable was also tested, by either including it or not including it in the modelling process. See Table 1 for a summary of the combinations of inputs.

Table 1: Description of the combinations of the input data assessed in the modelling process.
Inputs to modelling Description
Number of years of data 3 to 7 years
Number of lagged years for predictor variables 0, 1, and 2 lagged years
Inclusion of previous year’s value for outcome variable True or false
Number of different outcome variables modelled 5
Modelling methods explored GLM, RF (on outcome), RF (on change from previous year)

In all, nearly 450 different models were built and assessed. In reality, some of the time series for the outcome variables were not long enough to test for all variations of three to seven years, so fewer than 450 models were built.

Model selection

Each of the outcome variables are a proportion, so they have a scale of between 0 and 1. Despite the consistent scale, the distribution between within the scales of the outcome variables differs. MAPE was used (as described in Model evalaution) as a comparable metric across the different models.

Baseline comparison

As described in Model evalaution, two baseline naive methods of estimating future performance were performed and compared with the more sophisticated methods to contextualise the performance. The two methods chosen to compare with were selected because of their simplicity, but also because these are typical methods teams might employ for predicting future performance:

  • Next year’s performance is equal to the previous year’s performance
  • Next year’s performance is equal to a linear extrapolation based on the past 3 years’ performance

Important variables

The tables in this section show, for each modelling method and outcome, the most important variables selected by the models for the models with the lowest MAPE.

The MAPE for this model is 20.475

Variable Importance
lag_1_Proportion of incomplete pathways greater than 52 weeks from referral 0.0091642
lag_1_A and E and Ward Liaison mental health services spend (£k) per population 0.0058973
lag_1_Palliative care: QOF prevalence (all ages) 0.0022936
lag_1_Primary care workforce (FTEs) per 100,000 population (GP) 0.0020707
lag_1_A and E Full Time Equivalent (FTE) for Consultants (per 100,000 population) 0.0016539
lag_1_Primary care workforce (FTEs) per 100,000 population (Admin/Non-Clinical) 0.0016059
Primary care workforce (FTEs) per 100,000 population (GP) 0.0014976
Improving Access to Psychological Therapies spend (£k) per population 0.0013759
A and E Full Time Equivalent (FTE) for Doctors (excluding Consultants) (per 100,000 population) 0.0013757
A and E and Ward Liaison mental health services spend (£k) per population 0.0013093
Proportion of available beds that are occupied (General & Acute - day) 0.0011722
Asthma: QOF prevalence (all ages) 0.0011529
lag_1_Proportion of available beds that are occupied (General & Acute - overnight) 0.0011486
Diabetes mellitus: QOF prevalence (17+ yrs) 0.0010252
Children and Young People Mental Health spend - excluding learning disabilities and eating disorders per population 0.0009096
lag_1_Diabetes mellitus: QOF prevalence (17+ yrs) 0.0009069
lag_1_Children and Young People Mental Health spend - excluding learning disabilities and eating disorders per population 0.0008951
Proportion of population in age band (30-59) 0.0008580
Total beds per 1,000 60+ yrs (General & Acute - overnight) 0.0008316
lag_1_Stroke and transient ischaemic attack: QOF prevalence (all ages) 0.0008165

The MAPE for this model is 8.129

Variable Importance
lag_1_Proportion of suspected cancer or referral to first definitive treatment that are longer than 62 days wait 0.7524167
Obesity: QOF prevalence (18+ yrs) 0.0219428
Proportion of available beds that are occupied (General & Acute - overnight) 0.0173351
Mean proportion of beds that contain a patient with confirmed COVID 0.0100251
FTE days sick per 10,000 population (acute) 0.0083464
Diabetes mellitus: QOF prevalence (17+ yrs) 0.0076584
A and E Full Time Equivalent (FTE) for Consultants (per 100,000 population) 0.0075058
FTE days sick per 10,000 population (mental health) 0.0060682
Children and Young People Mental Health spend - excluding learning disabilities and eating disorders per population 0.0057620
A and E and Ward Liaison mental health services spend (£k) per population 0.0053697
Mental Health total spend per population 0.0052486
Adult community crisis care spend (£k) per population 0.0050600
Early Intervention in Psychosis spend (£k) per population 0.0046640
Total beds per 1,000 60+ yrs (General & Acute - overnight) 0.0044280
CKD: QOF prevalence (18+ yrs) 0.0043783
A and E Full Time Equivalent (FTE) for Nurses (per 100,000 population) 0.0043757
Proportion of available beds that are occupied (Maternity - overnight) 0.0037881
A and E Full Time Equivalent (FTE) for Doctors (excluding Consultants) (per 100,000 population) 0.0037671
Children and Young People Mental Health spend - eating disorders per population 0.0037593
Proportion of resident population in national deprivation quintile (Most deprived) - IMD 2019 0.0035042

The MAPE for this model is 13.398

Variable Importance
lag_1_Proportion of A&E attendances with greater than 4 hours wait (Type 1 Departments - Major A&E) 1.1643322
Obesity: QOF prevalence (18+ yrs) 0.2898131
Proportion of available beds that are occupied (General & Acute - overnight) 0.2409319
lag_1_Proportion of available beds that are occupied (General & Acute - overnight) 0.0248846
lag_1_Adult community crisis care spend (£k) per population 0.0151071
Mean proportion of beds that contain a patient with confirmed COVID 0.0140509
lag_1_Proportion of available beds that are occupied (General & Acute - day) 0.0123830
FTE days available per 100 population (acute) 0.0110055
Workforce FTEs per 100,000 population (Total - Acute) 0.0095914
Adult community crisis care spend (£k) per population 0.0093266
lag_1_FTE days sick per 10,000 population (acute) 0.0093117
Children and Young People Mental Health spend - eating disorders per population 0.0087977
lag_1_Obesity: QOF prevalence (18+ yrs) 0.0077410
lag_1_Osteoporosis: QOF prevalence (50+ yrs) 0.0067752
lag_1_Palliative care: QOF prevalence (all ages) 0.0065141
Mental health spend as a proportion of overall allocation 0.0060083
Proportion of available beds that are occupied (Mental Illness - overnight) 0.0054570
Primary care workforce (FTEs) per 100,000 population (GP) 0.0054178
CKD: QOF prevalence (18+ yrs) 0.0051650
Diabetes mellitus: QOF prevalence (17+ yrs) 0.0051542

The MAPE for this model is 11.918

Variable Importance
lag_1_Proportion of attended GP appointments (over 4 weeks wait time) 0.0211869
lag_1_Obesity: QOF prevalence (18+ yrs) 0.0017727
lag_1_Proportion of available beds that are occupied (General & Acute - overnight) 0.0007372
lag_2_Proportion of available beds that are occupied (General & Acute - overnight) 0.0006419
Adult community crisis care spend (£k) per population 0.0002177
Dementia: QOF prevalence (all ages) 0.0001895
lag_2_Primary care workforce (FTEs) per 100,000 population (Direct Patient Care) 0.0001750
Rheumatoid arthritis: QOF prevalence (16+ yrs) 0.0001530
lag_1_Primary care workforce (FTEs) per 100,000 population (Direct Patient Care) 0.0001277
lag_2_Proportion of available beds that are occupied (General & Acute - day) 0.0001253
Proportion of available beds that are occupied (General & Acute - overnight) 0.0000947
lag_2_Total beds per 1,000 60+ yrs (General & Acute - overnight) 0.0000829
Obesity: QOF prevalence (18+ yrs) 0.0000752
lag_2_Adult community crisis care spend (£k) per population 0.0000731
lag_1_Dementia: QOF prevalence (all ages) 0.0000676
Osteoporosis: QOF prevalence (50+ yrs) 0.0000650
lag_1_CKD: QOF prevalence (18+ yrs) 0.0000627
lag_1_Asthma: QOF prevalence (all ages) 0.0000617
lag_1_Adult community crisis care spend (£k) per population 0.0000570
Hypertension: QOF prevalence (all ages) 0.0000526

The MAPE for this model is 7.986

Variable Importance
lag_1_Proportion of incomplete pathways greater than 18 weeks from referral 0.2688382
Proportion of available beds that are occupied (General & Acute - overnight) 0.0201106
Diabetes mellitus: QOF prevalence (17+ yrs) 0.0162031
Proportion of available beds that are occupied (Mental Illness - overnight) 0.0114002
Obesity: QOF prevalence (18+ yrs) 0.0107161
lag_1_Diabetes mellitus: QOF prevalence (17+ yrs) 0.0099912
lag_1_Proportion of available beds that are occupied (General & Acute - overnight) 0.0097446
lag_1_Total beds per 1,000 60+ yrs (General & Acute - overnight) 0.0087030
lag_1_Proportion of available beds that are occupied (Mental Illness - overnight) 0.0073112
lag_1_Obesity: QOF prevalence (18+ yrs) 0.0053038
lag_1_A and E and Ward Liaison mental health services spend (£k) per population 0.0043959
lag_1_FTE days sick per 10,000 population (acute) 0.0036880
lag_1_Children and Young People Mental Health spend - eating disorders per population 0.0036461
% active smokers (GPPS) 0.0032558
Total beds per 1,000 60+ yrs (General & Acute - overnight) 0.0027749
lag_1_A and E Full Time Equivalent (FTE) for Care Providers (per 100,000 population) 0.0023889
Palliative care: QOF prevalence (all ages) 0.0023783
Mean proportion of beds that contain a patient with confirmed COVID 0.0021969
Mental health spend as a proportion of overall allocation 0.0021041
Osteoporosis: QOF prevalence (50+ yrs) 0.0021019

The MAPE for this model is 20.406

Variable Importance
Mental health spend as a proportion of overall allocation 0.0058820
COPD: QOF prevalence (all ages) 0.0054456
Obesity: QOF prevalence (18+ yrs) 0.0048128
Proportion of available beds that are occupied (General & Acute - overnight) 0.0036511
Proportion of resident population in national deprivation quintile (Most deprived) - IMD 2019 0.0035475
CKD: QOF prevalence (18+ yrs) 0.0029236
Hypertension: QOF prevalence (all ages) 0.0028695
Epilepsy: QOF prevalence (18+ yrs) 0.0024923
Proportion of population in age band (30-59) 0.0021667
% active smokers (GPPS) 0.0019919
A and E and Ward Liaison mental health services spend (£k) per population 0.0016337
Proportion of resident population in national deprivation quintile (Least deprived quintile) - IMD 2019 0.0015969
Proportion of population in age band (0-29) 0.0015252
Peripheral arterial disease: QOF prevalence (all ages) 0.0012288
Asthma: QOF prevalence (all ages) 0.0011564
Mental Health total spend per population 0.0011356
Early Intervention in Psychosis spend (£k) per population 0.0011276
A and E Full Time Equivalent (FTE) for Doctors (excluding Consultants) (per 100,000 population) 0.0010177
Total beds per 1,000 60+ yrs (General & Acute - overnight) 0.0010051
Stroke and transient ischaemic attack: QOF prevalence (all ages) 0.0009966

The MAPE for this model is 6.599

Variable Importance
lag_2_Dementia: QOF prevalence (all ages) 0.0245797
lag_2_Obesity: QOF prevalence (18+ yrs) 0.0230728
lag_2_Proportion of available beds that are occupied (General & Acute - day) 0.0174987
lag_2_Proportion of available beds that are occupied (General & Acute - overnight) 0.0074086
lag_2_Asthma: QOF prevalence (all ages) 0.0062151
% active smokers (GPPS) 0.0024263
lag_1_Dementia: QOF prevalence (all ages) 0.0020730
lag_1_Cancer: QOF prevalence (all ages) 0.0019017
lag_2_Primary care workforce (FTEs) per 100,000 population (GP) 0.0018205
lag_1_Palliative care: QOF prevalence (all ages) 0.0017027
lag_2_Osteoporosis: QOF prevalence (50+ yrs) 0.0016949
lag_2_Mental health spend as a proportion of overall allocation 0.0016423
lag_1_Mental Health total spend per population 0.0015775
lag_1_Atrial fibrillation: QOF prevalence (all ages) 0.0015507
lag_2_Proportion of available beds that are occupied (Mental Illness - overnight) 0.0013287
Peripheral arterial disease: QOF prevalence (all ages) 0.0012338
lag_1_Hypertension: QOF prevalence (all ages) 0.0012040
lag_2_Palliative care: QOF prevalence (all ages) 0.0011607
lag_2_Proportion of available beds that are occupied (Maternity - overnight) 0.0011537
lag_1_Proportion of available beds that are occupied (General & Acute - overnight) 0.0010800

The MAPE for this model is 11.562

Variable Importance
Proportion of available beds that are occupied (General & Acute - overnight) 0.0792941
Obesity: QOF prevalence (18+ yrs) 0.0649249
Proportion of available beds that are occupied (General & Acute - day) 0.0454916
Asthma: QOF prevalence (all ages) 0.0448135
Dementia: QOF prevalence (all ages) 0.0394395
Mean proportion of beds that contain a patient with confirmed COVID 0.0353722
FTE days sick per 10,000 population (mental health) 0.0349679
Osteoporosis: QOF prevalence (50+ yrs) 0.0260058
Total beds per 1,000 60+ yrs (General & Acute - overnight) 0.0229201
FTE days sick per 10,000 population (acute) 0.0226379
Hypertension: QOF prevalence (all ages) 0.0175096
Proportion of available beds that are occupied (Mental Illness - overnight) 0.0155020
Mental health spend as a proportion of overall allocation 0.0153613
Palliative care: QOF prevalence (all ages) 0.0149768
CKD: QOF prevalence (18+ yrs) 0.0134940
Rheumatoid arthritis: QOF prevalence (16+ yrs) 0.0131765
Diabetes mellitus: QOF prevalence (17+ yrs) 0.0109324
COPD: QOF prevalence (all ages) 0.0105366
Atrial fibrillation: QOF prevalence (all ages) 0.0101031
Improving Access to Psychological Therapies spend (£k) per population 0.0083710

The MAPE for this model is 7.871

Variable Importance
lag_1_Dementia: QOF prevalence (all ages) 0.0003675
lag_1_Obesity: QOF prevalence (18+ yrs) 0.0003217
Hypertension: QOF prevalence (all ages) 0.0003208
lag_1_Asthma: QOF prevalence (all ages) 0.0003094
lag_1_Atrial fibrillation: QOF prevalence (all ages) 0.0003066
lag_1_Proportion of available beds that are occupied (General & Acute - overnight) 0.0002377
lag_1_Proportion of attended GP appointments (over 4 weeks wait time) 0.0002221
lag_1_Hypertension: QOF prevalence (all ages) 0.0002021
lag_1_Diabetes mellitus: QOF prevalence (17+ yrs) 0.0001802
lag_1_Proportion of available beds that are occupied (General & Acute - day) 0.0001709
lag_1_Total beds per 1,000 60+ yrs (General & Acute - overnight) 0.0001537
lag_1_Osteoporosis: QOF prevalence (50+ yrs) 0.0001134
lag_1_Proportion of available beds that are occupied (Mental Illness - overnight) 0.0001125
Obesity: QOF prevalence (18+ yrs) 0.0000947
Mental health spend as a proportion of overall allocation 0.0000930
Proportion of available beds that are occupied (General & Acute - overnight) 0.0000887
lag_1_Heart failure: QOF prevalence (all ages) 0.0000655
Proportion of resident population in national deprivation quintile (Most deprived) - IMD 2019 0.0000590
COPD: QOF prevalence (all ages) 0.0000567
lag_1_Mental health spend as a proportion of overall allocation 0.0000541

The MAPE for this model is 6.236

Variable Importance
Asthma: QOF prevalence (all ages) 0.2676618
Obesity: QOF prevalence (18+ yrs) 0.1511454
Proportion of available beds that are occupied (General & Acute - overnight) 0.1113777
Dementia: QOF prevalence (all ages) 0.0689070
lag_1_Asthma: QOF prevalence (all ages) 0.0450871
lag_1_Osteoporosis: QOF prevalence (50+ yrs) 0.0224910
lag_1_FTE days sick per 10,000 population (mental health) 0.0069164
lag_1_Obesity: QOF prevalence (18+ yrs) 0.0040761
lag_2_Proportion of available beds that are occupied (General & Acute - day) 0.0035922
lag_1_Proportion of available beds that are occupied (General & Acute - overnight) 0.0032268
FTE days available per 100 population (mental health) 0.0030448
Proportion of available beds that are occupied (Mental Illness - overnight) 0.0027784
lag_2_Hypertension: QOF prevalence (all ages) 0.0027642
lag_2_Adult community crisis care spend (£k) per population 0.0025911
Peripheral arterial disease: QOF prevalence (all ages) 0.0025122
COPD: QOF prevalence (all ages) 0.0024921
lag_2_Children and Young People Mental Health spend - eating disorders per population 0.0024679
lag_2_Palliative care: QOF prevalence (all ages) 0.0022773
lag_2_Proportion of population in age band (0-29) 0.0022134
lag_1_FTE days sick per 10,000 population (acute) 0.0021042

The MAPE for this model is 23.992

Variable Importance Sign
Obesity: QOF prevalence (18+ yrs) 0.1970321 NEG
lag_1_Proportion of incomplete pathways greater than 52 weeks from referral 0.1290261 POS
Primary care workforce (FTEs) per 100,000 population (Admin/Non-Clinical) 0.1033912 POS
Primary care workforce (FTEs) per 100,000 population (GP) 0.1007553 POS
Diabetes mellitus: QOF prevalence (17+ yrs) 0.0798783 POS
Proportion of population in age band (30-59) 0.0688582 NEG
Proportion of available beds that are occupied (General & Acute - day) 0.0508516 NEG
Proportion of available beds that are occupied (Maternity - overnight) 0.0486288 POS
Children and Young People Mental Health spend - excluding learning disabilities and eating disorders per population 0.0404209 POS
FTE days available per 100 population (mental health) 0.0396462 NEG
A and E Full Time Equivalent (FTE) for Doctors (excluding Consultants) (per 100,000 population) 0.0335540 NEG
Proportion of available beds that are occupied (General & Acute - overnight) 0.0237718 NEG
Palliative care: QOF prevalence (all ages) 0.0212065 POS
A and E Full Time Equivalent (FTE) for Consultants (per 100,000 population) 0.0095311 NEG

The MAPE for this model is 6.22

Variable Importance Sign
lag_1_Proportion of suspected cancer or referral to first definitive treatment that are longer than 62 days wait 0.2959267 POS
Proportion of available beds that are occupied (General & Acute - overnight) 0.0759739 POS
Mean proportion of beds that contain a patient with confirmed COVID 0.0682840 POS
Adult community crisis care spend (£k) per population 0.0185589 NEG
Primary care workforce (FTEs) per 100,000 population (Direct Patient Care) 0.0185479 POS
CKD: QOF prevalence (18+ yrs) 0.0179939 POS
% active smokers (GPPS) 0.0172982 POS
FTE days sick per 10,000 population (acute) 0.0157005 POS
Mental health spend as a proportion of overall allocation 0.0102860 POS
Osteoporosis: QOF prevalence (50+ yrs) 0.0102827 NEG
Primary care workforce (FTEs) per 100,000 population (GP) 0.0041997 NEG
Diabetes mellitus: QOF prevalence (17+ yrs) 0.0036469 POS
FTE days sick per 10,000 population (mental health) 0.0036074 POS
A and E Full Time Equivalent (FTE) for Care Providers (per 100,000 population) 0.0025265 POS
Proportion of available beds that are occupied (Mental Illness - overnight) 0.0023008 POS
A and E Full Time Equivalent (FTE) for Nurses (per 100,000 population) 0.0008777 NEG
Proportion of available beds that are occupied (General & Acute - day) 0.0001072 NEG

The MAPE for this model is 15.54

Variable Importance Sign
lag_1_Proportion of A&E attendances with greater than 4 hours wait (Type 1 Departments - Major A&E) 0.3784697 POS
Proportion of available beds that are occupied (General & Acute - overnight) 0.1559352 POS
lag_2_Proportion of available beds that are occupied (General & Acute - overnight) 0.0795808 NEG
lag_1_Proportion of available beds that are occupied (General & Acute - day) 0.0641298 NEG
lag_1_Obesity: QOF prevalence (18+ yrs) 0.0613511 NEG
lag_1_Proportion of available beds that are occupied (General & Acute - overnight) 0.0551466 NEG
FTE days sick per 10,000 population (acute) 0.0547416 POS
Obesity: QOF prevalence (18+ yrs) 0.0396580 POS
lag_2_Adult community crisis care spend (£k) per population 0.0266187 NEG
Mean proportion of beds that contain a patient with confirmed COVID 0.0191579 POS
Primary care workforce (FTEs) per 100,000 population (GP) 0.0187365 POS
A and E Full Time Equivalent (FTE) for Care Providers (per 100,000 population) 0.0157239 POS
Mental health spend as a proportion of overall allocation 0.0133069 NEG
lag_1_Primary care workforce (FTEs) per 100,000 population (GP) 0.0121372 POS
Early Intervention in Psychosis spend (£k) per population 0.0118444 POS
CKD: QOF prevalence (18+ yrs) 0.0110565 POS
Children and Young People Mental Health spend - excluding learning disabilities and eating disorders per population 0.0109340 POS
lag_1_Palliative care: QOF prevalence (all ages) 0.0079181 NEG
lag_1_Osteoporosis: QOF prevalence (50+ yrs) 0.0077961 NEG
A and E Full Time Equivalent (FTE) for Doctors (excluding Consultants) (per 100,000 population) 0.0019828 NEG

The MAPE for this model is 6.449

Variable Importance Sign
lag_1_Proportion of attended GP appointments (over 4 weeks wait time) 0.3277719 POS
lag_2_Proportion of available beds that are occupied (General & Acute - overnight) 0.0744769 NEG
lag_1_Proportion of available beds that are occupied (General & Acute - overnight) 0.0740342 POS
Asthma: QOF prevalence (all ages) 0.0695301 POS
lag_1_Obesity: QOF prevalence (18+ yrs) 0.0410142 POS
lag_2_Obesity: QOF prevalence (18+ yrs) 0.0374213 NEG
lag_2_Primary care workforce (FTEs) per 100,000 population (GP) 0.0263894 NEG
lag_2_CKD: QOF prevalence (18+ yrs) 0.0261564 NEG
Adult community crisis care spend (£k) per population 0.0248809 NEG
lag_1_Proportion of available beds that are occupied (Mental Illness - overnight) 0.0241289 POS
Proportion of population in age band (30-59) 0.0237368 NEG
Proportion of resident population in national deprivation quintile (Least deprived quintile) - IMD 2019 0.0230094 POS
A and E Full Time Equivalent (FTE) for Care Providers (per 100,000 population) 0.0213554 POS
Primary care workforce (FTEs) per 100,000 population (Nurses) 0.0212858 POS
Children and Young People Mental Health spend - excluding learning disabilities and eating disorders per population 0.0206432 NEG
Proportion of available beds that are occupied (General & Acute - overnight) 0.0200274 POS
Dementia: QOF prevalence (all ages) 0.0135149 POS
lag_1_Primary care workforce (FTEs) per 100,000 population (Nurses) 0.0133500 POS
lag_2_Proportion of population in age band (30-59) 0.0131589 NEG
Early Intervention in Psychosis spend (£k) per population 0.0118832 POS

The MAPE for this model is 7.718

Variable Importance Sign
lag_1_Proportion of incomplete pathways greater than 18 weeks from referral 0.1876863 POS
lag_1_Proportion of available beds that are occupied (General & Acute - overnight) 0.0236748 POS
lag_2_Proportion of available beds that are occupied (General & Acute - overnight) 0.0114581 NEG
Proportion of available beds that are occupied (Mental Illness - overnight) 0.0073651 POS
Proportion of available beds that are occupied (Maternity - overnight) 0.0004970 POS

Summary of best models

Table 2 presents the combinations of the inputs to the modelling that created the lowest MAPE for each outcome, along with the baseline comparisons.

Table 2: By outcome, the best combination of input data, along with the MAPE score when comparing the predicted outcome with the observed outcome on the data held back for testing.
Target variable Model type Target variable type Number of training years Number of lagged years Include previous year’s outcome value MAPE (test data) MAPE (baseline - last year) MAPE (baseline - linear 3 years)
Proportion of incomplete pathways greater than 52 weeks from referral Random forest difference from previous 3 0 TRUE 20.406 33.064 51.775
Proportion of suspected cancer or referral to first definitive treatment that are longer than 62 days wait Generalised linear model proportion 4 0 TRUE 6.220 15.185 12.326
Proportion of A&E attendances with greater than 4 hours wait (Type 1 Departments - Major A&E) Random forest difference from previous 4 0 FALSE 11.562 31.590 36.270
Proportion of attended GP appointments (over 4 weeks wait time) Generalised linear model proportion 3 2 TRUE 6.449 18.809 49.468
Proportion of incomplete pathways greater than 18 weeks from referral Random forest difference from previous 4 2 FALSE 6.236 20.141 23.163