The Economic Burden of Hip Fractures Among Elderly Patients in Ireland: A Combined Perspective of System Dynamics and Machine Learning

The 34th International Conference of the System Dynamics Society
The Economic Burden of Hip Fractures among

Elderly Patients in Ireland: A Combined Perspective
of System Dynamics and Machine Learning
Mahmoud Elbattah, Owen Molloy
m.elbattah1@nuigalway.ie
Challenge to Healthcare: Population Ageing
Source : Health Service Executive. Annual Report and Financial Statements, 2014. 2
Our Focus: Hip Fracture Care in Ireland
• A good exemplar of elderly healthcare.
• Exponentially increasing with age.1
• Identified as one of the most serious injuries resulting in lengthy

hospital admissions and high costs.2
• Availability of empirical data through the Irish Hip Fracture Database

(IHFD).
Sources :1 Gullberg, B., Johnell, O. and Kanis, J.A., 1997. World-wide projections for hip fracture. Osteoporosis international, 7(5), pp.407-413.
2http://www.hse.ie/eng/services/publications/olderpeople/Executive_Summary_Strategy_to_Prevent_Falls_and_Fractures_in_Ireland%E2%80%9 9s_Ageing_Po
pulation.pdf 3
Question of Interest
• With the growing trend of population ageing, how could be the

potential economic burden of elderly hip-fracture patients on the
healthcare system in Ireland over the next 10 years?
Given that:
Cost of Treatment =
(ED Cost) + (Hospital Inpatient Cost) +
(Outpatient Visits Cost) + (Long-Stay Care Cost)
4
Related Questions
Q1) How to predict the inpatient length of stay in acute facilities?
Q2) How to predict the discharge destination for a hip-fracture

patient?
5
Our Approach: Integrating Simulation Modeling with

Machine Learning
Patient-Focused Perspective Population-Driven Perspective
Machine Learning
+ Simulation Modeling
Predict LOS and Modeling Projected

Destination Discharge Flow of Elderly Patients
Data-Driven Knowledge Domain Knowledge

Methodology Overview
7
Sources of Data
• Irish Hip Fracture Database (IHFD). (Year 2013)
• Population projections from the Central Statistics Office (CSO).
• Additional population statistics with respect to CHOs from the Health

Intelligence Department.
8
The Initial SD Model
Long-Stay Care
Hip Fracture Rate for Discharged
Elderly Males
Discharge Fraction
Potential Male Total Discharged
+ + Patients
Patients
+
New Male Cases
+ InHospital
+ Home-Discharged
Total Elderly
Population New Female R
+ Cases
+ Return Patients +
+ +
+
Potential Female
Patients
Recurrence
Hip Fracture Rate for Fraction
Elderly Females
9
Disaggregating the Model

Potential Male
Patients-CHO1
New Male InHospital
CHO1 Elderly Cases-CHO1
Population
(CHO1) Discharged
Patients- CHO1
New Female
Cases-CHO1
Potential Female
Patients-CHO1
Potential Male
Patients-CHO2
New Male InHospital
Cases-CHO2 (CHO2) Discharged
CHO2 Elderly
Population Patients-CHO2
New Female
Cases-CHO2
Potential Female
Patients-CHO2
Potential Male
Patients-CHO3
New Male InHospital
CHO3 Elderly Cases-CHO3 (CHO3) Discharged
New Female
Cases-CHO3
Potential Female
Patients-CHO3
Hip Fracture Rate

Potential Male
for Elderly Males Patients-CHO4
Total Elderly New Male InHospital
Population CHO4 Elderly Cases-CHO4 (CHO4) Discharged
New Female
Cases-CHO4
Potential Female
Patients-CHO4
Potential Male
Patients-CHO5
New Male InHospital
New Female
Cases-CHO5
Potential Female
Patients-CHO5
Potential Male Total Discharged

Hip Fracture Rate for Patients-CHO6 New Male InHospital Patients
Cases-CHO6 (CHO6) Discharged
Elderly Females CHO6 Elderly Patients-CHO6
Population New Female
Cases-CHO6
Potential Female
Patients-CHO6
Potential Male
Patients-CHO7
New Male InHospital
New Female
Cases-CHO7
Potential Female
Patients-CHO7
Potential Male
Patients-CHO8
New Male InHospital
CHO8 Elderly Cases-CHO8
Population (CHO8) Discharged
Patients-CHO8
New Female
Cases-CHO8
Potential Female
Patients-CHO8
Potential Male
Patients-CHO9
New Male InHospital
New Female
Cases-CHO9
Potential Female
Patients-CHO9
10
Generation of Patients
Counts of patients generated per CHO over 50 simulation experiments.
Community Health No. of Simulation-Generated Patients
Organisation (CHO)
CHO1 151,850
CHO2 169,550
CHO3 142,450
CHO4 247,750
CHO5 187,050
CHO6 140,750
CHO7 191,900
CHO8 187,050
CHO9 180,650
11
Machine Learning Models

• Regression Model -> Length of Stay.
• Classification Model -> Discharge destination.
• Machine learning algorithm: Random Forests.
12
Machine Learning Models (cont’d)

Average 10-fold cross-validation accuracy of the LOS predictor
Relative Absolute Error Relative Squared Error Coefficient of Determination
≈0.26 ≈0.17 ≈0.83
Average 10-fold cross-validation accuracies of discharge destination classifier.

13
Calculation of Cost
Cost of Treatment = (ED Cost) + (Hospital Inpatient Cost) + (Outpatient

Visits Cost) + (Long-Stay Care Cost)
• Information on costs was acquired from the report “The economic

costs of falls and fractures in people aged 65 and over in Ireland”.1
1Gannon, B., O’Shea, E. and Hudson, E., 2007. The economic costs of falls and fractures in people aged 65 and over in Ireland. Irish Centre for Social
Gerontology, Galway. 14
Results: Predicted Cost in 10 Years
15
Results: Predicted Costs in CHOs

200,000,000
175,000,000
150,000,000
Average Accumulative
125,000,000
100,000,000
75,000,000
50,000,000
25,000,000
0
CHO1 CHO2 CHO3 CHO4 CHO5 CHO6 CHO7 CHO8 CHO9
Home-Discharged Long-Stay Care Discharged
16
Visualising Predicted Costs in CHOs
Heatmap: Overall predicted cost within every CHO. 17

Study Limitations
• Only public acute hospitals were considered.
• The IHFD dataset did not evenly represent the 9 CHOs.
• The dataset covered only a single year (2013).
• The rate of hip fractures was assumed as a constant over the

simulated interval, however it might increase or decrease in reality.
18
Study Limitations (cont’d)

• In-hospital cost of the patients aged 60-64 were considered the same
as 65-69.
• The study did not consider other potential costs such as the
ambulance costs.
• The study did not consider the indirect costs such as the quality of
life.
• The study did not distinguish between the patients who are
discharged to long-stay nursing homes and rehabilitation institutions.
19
Discussion
• Why not use Simulation Modeling alone?
• Why not use Machine Learning alone?
20
Discussion (cont’d)
21
Studies Integrating Simulation & ML
22
Studies Integrating Simulation & ML
23
Acknowledgements
• System Dynamics Society.
• National Office of Clinical Audit (NOCA), Ireland.
24
THANK YOU!
Mahmoud Elbattah
m.elbattah1@nuigalway.ie

The Economic Burden of Hip Fractures Among Elderly Patients in Ireland: A Combined Perspective of System Dynamics and Machine Learning

Hochgeladen von

Dokumentinformationen

Originaltitel

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

The Economic Burden of Hip Fractures Among Elderly Patients in Ireland: A Combined Perspective of System Dynamics and Machine Learning

Hochgeladen von

Copyright:

Verfügbare Formate

The 34th International Conference of the System Dynamics Society

The Economic Burden of Hip Fractures among

Challenge to Healthcare: Population Ageing

Our Focus: Hip Fracture Care in Ireland

• A good exemplar of elderly healthcare.

• Exponentially increasing with age.1

• Identified as one of the most serious injuries resulting in lengthy

• Availability of empirical data through the Irish Hip Fracture Database

• With the growing trend of population ageing, how could be the

Q2) How to predict the discharge destination for a hip-fracture

Our Approach: Integrating Simulation Modeling with

Patient-Focused Perspective Population-Driven Perspective

Predict LOS and Modeling Projected

Data-Driven Knowledge Domain Knowledge

• Population projections from the Central Statistics Office (CSO).

• Additional population statistics with respect to CHOs from the Health

The Initial SD Model

Disaggregating the Model

Hip Fracture Rate

Potential Male Total Discharged

Machine Learning Models

• Classification Model -> Discharge destination.

• Machine learning algorithm: Random Forests.

Machine Learning Models (cont’d)

Average 10-fold cross-validation accuracies of discharge destination classifier.

Cost of Treatment = (ED Cost) + (Hospital Inpatient Cost) + (Outpatient

• Information on costs was acquired from the report “The economic

Results: Predicted Cost in 10 Years

Results: Predicted Costs in CHOs

Visualising Predicted Costs in CHOs

Heatmap: Overall predicted cost within every CHO. 17

• The IHFD dataset did not evenly represent the 9 CHOs.

• The dataset covered only a single year (2013).

• The rate of hip fractures was assumed as a constant over the

Study Limitations (cont’d)

• Why not use Machine Learning alone?

Studies Integrating Simulation & ML

Studies Integrating Simulation & ML

• System Dynamics Society.

• National Office of Clinical Audit (NOCA), Ireland.

Das könnte Ihnen auch gefallen