Proposal Defence Presentation.pdf

Smart Medical Charge Calculation System First Evaluation

Hendy Risdianto Wijaya PMB173018

Outline I. Introduction • • • • •

Problems Background Problems Statement Research Hypothesis Research Objective Scope of Study

II. Literature Review • • • • •

Health Financing DRG INA-CBGs Clinical Pathway Artificial Intelligent

III. Research Methodology • • • •

Proposed Method Assumption Research Design and Procedure Research Planning

IV. Expected Result V. Conclusion

Introduction – Problems background  Universal Health Coverage (UHC) UHC initiated by WHO is a parameter for government to guarantee health access to everybody.  Universal Health Coverage (UHC) implemented in Indonesia through JKN (Jaminan Kesehatan Nasional / National Health Security) program is deploying the casemix system called INA-CBGs (Indonesia Case Based Groups).  The program is managed by BPJS Health (Badan Penyelenggara Jaminan Sosial in Health / Social Insurance Administration Organization in Health) and targeting 2019 to reach 100% UHC.  The INA-CBGs system which is based on static formulas has caused un-satisfaction to patient, doctors, and hospital in Indonesia. These problems are primarily caused inaccurate input parameters, calculation method, unsuitable clinical pathway and lack of databases to identify severity.

Introduction – Problems Statement Existing medical cost calculation methods are still inaccurate, caused by:  Inaccurate input parameters.  The input used for the current system is costing and coding of hospitals  Only 157 hospitals from 2218 Indonesia hospitals contributed.

 Errors in clinical coding.  Most Doctors do not understand the classification of diseases based on ICD10 and ICD9, Coder will help doctor to define that.  Coder incorrectly defines coding from a doctor's diagnosis

 Unsuitable clinical pathway in hospital.  BPJS does not provide instructions for using clinical pathways.  Many hospital claims are not accepted because the clinical pathway is not appropriate.

 The income of health workers must be balanced with service.  Hospitals cannot plan hospital finances.

Introduction – Research Hypothesis  The appropriate medical cost calculation system will be able to overcome data inaccuracies, accuracy in calculating unit costs and inaccurate classification of diseases  The appropriate medical cost calculation system will ensure the satisfaction of health stakeholders (doctor, patient and hospital).  The appropriate medical cost calculation system will be able to predict the health costs of patients, hospitals and government .

Introduction – Research Objective 1. To evaluate existing medical charge calculation system. 2. To develop new system for calculating medical charge using a combination of artificial intelligence methods. 3. To evaluate new medical charge calculation in term:  The payment fairness (patient and health worker)  Accuracy of disease classification  Medical charge accuracy  Health care cost prediction accuracy.

Introduction – Scope of study Workflow Scope • • • •

Literature studies (journal, expert interview and survey) Benchmarking (DRG concept and existing INA-CBG tariff calculation) Data collection (expert, doctor, patient and hospital finance) Data analysis, system modeling, simulation (using RapidMiner Studio and MATLAB) and testing (medical charge calculation and disease classification).

Material Scope • Focus on the highest cause of death and catastrophic illness in Indonesian such as Stroke, Cardiovascular Disease, Cancer, and Kidney Failure. • Focus on medical charge calculation system in Indonesia, other countries also discuss for comparison.

Literature study – Health Financing Health Cost  Health Costs in consumer side  Resource: Tax Based Funding (Government Funding and Government Insurance), Community Based Health Funding (Global Fund), Private Sector (Private Funding and Private Insurance), Out of Pocket (OOP)  Payment Method:  Retrospective Payment (FFS = Fee for Service)  Prospective Payment (Casemix)  Health Costs in Provider side  Basic cost classification: Fixed Cost (FC), Variable Cost (VC), Semi Fixed Cost, Semi Variable Cost, Total Cost (TC), Annualized Fixed Cost (AFC).  Cost Type based on the effect on changes in production scale, based on the duration of use, based on the function and activity of cost sources, and based on unit-cost

Literature study – Health Financing Health Service Charge (tariff)  Propose of tariff calculation:  Determination of Cost Recovery Rates (CRR) and Cross Subsidies Cost.  Improve Service access and quality.  Reduce competitors, maximize revenue, minimize use, create corporate image.  Determination Process:  Full-cost pricing > Determine tariff according to unit costs plus profits.  Contracts and cost-plus > Hospital tariff can be determined based on contracts for insurance companies, or consumers who are members of one organization  Target rate of return pricing > This method is a modification of the full-cost method above, ex: tariff determined by directors must have a 10% profit.  Acceptance pricing > This technique is used if on the market there is one hospital that is considered a role model (market leader) in price and the price is referred to by other hospitals.  Pricing by looking at competitor pricing above / below competitors  Setting tariffs on government organizations (according to the government)

Literature study – DRG History  The most widely used casemix classification  Depend on their goal and purpose, as well as modality of application vary importantly.  Used to classify acute admitted care (not used for non-admitted care)  Classes defined by principal medical diagnosis, plus variables such as other diagnoses and procedures.  These variables are ‘cost-drivers’: They drive (predict) the cost of acute care (not good predictors to predict mental health). DRG’s part is updated regularly: • Updates of Cost Weights (CW) • Update of Casemix Index (CMI) • Update of Base Rates (HBR) • Adjustment Factor (aF)

: Every 2 years : Every year : Every year : Every year

Country

Literature study

DRG Comparison Diagnosis Related Groups  Developed based on acute health conditions  Uses mainly diagnosis and procedures in the classification system.  Implementing DRG: Grouper, Diagnosis, Procedure Impact of DRG Mainly for cost-containment:  create inventive for hospital to control/reduce the costs  reduce the LOS of patients and increase number of patient admissions > increase efficiency Problems:  Reduction in Length of Stay (up to 25%)  Low rate of growth of hospital costs.  Decrease in hospital profit margins. > Other effects: up-coding, effect on out-patients care cost.

Year of

Origin of system Based on US model and adapted to Australian clinical data Based on Nordic DRG, Danish-specific version developed Based on Yale system with modification and refinement Based on AR-DRG and adapted to German procedure codes

Name of Casemix

Severity Level

Extent of System use

Hospital payment scheme

AR-DRG

4

In-patient hospital care, Out-patient and emergency care (different PCS)

Payment per case/ DRG (47%) + retrospective reimbursement (48%)

Nordic DRG

2

All hospital activity

Prospective global budget (80%) + Payment per case/DRG (20%)

GHM

5

Acute in-patient care

Payment per case/DRG

G-DRG

not limited

All hospital activity

Global budget +Payment per case/DRG

Australia (45,46)

1993

Denmark

2002

France

1994

Germany (47–49)

2005

Italy

1994

Based on US system

CMS-DRG

Japan (26)

2003

Original system

DPC

Portugal

1987

Based on US system

CMS-DRG

2

Acute in-patient care

Prospective global budget based on DRG

Singapore (50,51)

1999

Based on AN-DRG

IR-DRG

3

Acute in-patient care

Global budget +Payment per case/DRG

Sweden

1991

Nordic DRG

2

UK (52)

1991

HRG

3

USA (53–55)

1983

Nordic-DRG based on HCFA-DRG Based on HCFA-DRG with much refinement HCFA-DRG (origin)*, + APR-DRG, MS-DRG, APG, etc

Mainly CMS-DRG

3

Indonesia (56–58)

2006

Based on UNU-CBG

INA-CBG

3

Malaysia (59,60)

2002

Based on UNU-CBG

My-DRG

3

Thailand (61,62)

1999

Based on HFCA-DRG, +AR-DRG

Thai-DRG

5

Philippine (63)

2010

Based on UNU-CBG

PH-DRG

1

2

In-patient hospital care extends to nursing homes Acute in-patient care

Global budget +Payment per case/DRG Payment per diem/DPC

Acute in-patient care + Psychiatric wards In-patient hospital care, Out-patient, emergency care, chronic illness In-patient hospital care, Out-patient, emergency care, chronic illness In-patient hospital care, Out-patient, sub-acute, chronic illness In-patient hospital care, Out-patient In-patient hospital care, emergency care, chronic illness

Global budget +Payment per case/DRG (55%) Global budget (30%) +Payment per case/DRG (70%)

Acute in-patient care

Payment per case/CBG

Payment per case/DRG

Payment per case/CBG Payment per case/CBG Payment per case/CBG

Literature study – INA-CBGs (existing tariff calculation)(1) INA-CBGs Tariff = Hospital Base rate x Cost Weight x aF Hospital Base Rate Hospital cost = Number of Hospital equivalent x CMI

Cost Weight = CMI=

∑(

#





#



)



Literature study – INA-CBGs (existing tariff calculation)(2)

Literature study – INA-CBGs (existing tariff calculation)(3) Re-classification MDC 12 : EYE & ADNEXA MDC 13 : ENT & MOUTH MDC 14 : RESPIRATORY MDC 29 : PSYCHIATRY

2016

2016

Tariff Instrument costing improvements

Re-classification Others MDC

2017 2018

Re-classification Finalization & Trial INA-Grouper

2017 2018

Tariff Collecting data costing

2018 2019

2018 2019

New Grouper & Tariff

Tariff Processing & Finalization of Tariff

System improvements and calculations from INA-CBGs (National Casemix Center MOH - 2019)

Literature study – Clinical Pathway  Clinical Pathway is a flow of patient service activities that are specific to a particular disease or action, from patient admission to patient discharge which is an integration of medical services, nursing services, pharmaceutical services and other health services.  Clinical pathways are very instrumental in patient recovery, length of stay and health financing.  Every hospital can define Clinical Pathways based on facilities and doctors they have.  To ensure service quality, there is a need for minimum and standard clinical pathways.

Literature study – Clinical Pathway for stroke

Literature study – Artificial Intelligent Fuzzy Logic (FL)

Artificially Neural Network (ANN)

Advantage

Solve non-linear and complex problems, solve problems that are nowhere related to each other

Solve non-linear and complex problems, solve problems that are nowhere related to each other

Disadvantage

membership classification must be clear

Can be over fitting with not accurate input so the pre-processing needs to be done.

Year of invention

1965

1943

Basic principle

Make definite decisions based on incorrect or ambiguous data (relating to fixed and approximate reasons)

Incorporate human thinking process to solve problems without mathematically modeling them

Application Type

Classification and clustering, decision support system

Supervised Learning, Clustering, Structured Prediction, Anomaly detection

Process

Development of membership functions and rules to relate them.

Involves learning algorithms and requires training data

Expression

Many-valued logic , nay real number between 0 Using hidden layer and weightage for each and 1 input.

Research Methodology – Proposed Model Risk Predictor Intake Life Stye Activities Life Style

Risk Factor Sign and Symptom

Ward Level Severity Level

Prev Diagnostic Clinical Pathway Company

Subsidies

Medical Charge

Out of pocket Salary/Income

Pay-ability Level Private Insurance Public Insurance

Government Health Expenditure

Human Capital Level Facilities Level

Private Investor

Medication Level

Unit Cost Level

Smart Medical Charge Calculation

Severity Level (Svl) = f(Rf, SS, PrevD) Pay-Ability Level (PAb) = f(Salary, Gov, Com) Unit Cost Level (UC) = f(HWorker,Facility,Drug) Medical Charge = f(Svl,PAb,UC)

Research Methodology – Assumption Clinical Pathway: • Clinical Pathway is the basic for cost calculation. • Clinical Pathway depends on patient severity, patient pay-ability, facilities, medication, and doctor. Hospital issue: • Hospital benefits range from 5-15% of the total cost depending on the cost item and must be determined by the government. • The costs of health workers have standard costs set by the government (Salary of doctors and nurses in hospitals based on education, work experience and achievement) • Hospital service charges will be adjusted to the patient's ability to pay. Severity Level: • Severity of disease depends on Risk of disease Predictor (doctor diagnosis on machine learning). • Severity levels are the basis for choosing a Clinical Pathway. Medical Charge • Unit Cost depends on facilities, medication and doctor. • This system focus on input to define medical charge. • Medical charge depend on Clinical Pathway that patient choose.

Research Methodology – Research Design and Procedure (1) 1.

Review and studies the existing medical charge calculation system to get new specifications of the calculation and develop it according to the user requirement existing calculation system.

2.

Develop a new method that is optimized through data collection for model parameters, simulations, and model tests.

Research Methodology – Research Design and Procedure (2) 3.

The optimal model obtained in the previous phase will be validated and evaluated. Tests were carried out in several hospitals.  Develop validation and evaluation method of Medical Charge System  Develop simulation model on the Medical Charge system.  Develop survey questioner to doctor, nurse, patients, and hospital directors in term satisfaction measurement.  Develop proposal to get ethical approval (for hospital) and obtain ethical approval  Conduct survey, interview, or data collection.  Analysis, report writing and paper writing (Publish)

Research Methodology – Research Planning No

Activities

1 Problem Identifications 2 Literature/comparison study 3 Review of existing tariffication system 4 Interview with expert 5 Data Analysis (Stroke Management Cost) 6 Develop System Design (Smart MC Calculation) 7 Tesis Writing / Report 1 8 Paper 1 9 Data Analysis for Parameter Model 1 (Risk Predictor) 10 Develop MODEL 1 (Risk Predictor Model) 11 Simulation and testing MODEL 1 (Risk Predictor) 12 Field Data collection 1a (Risk Predictor on Stroke) 13 Analysis MODEL 1 & Tesis Writing / Report 2 14 Submit and Publish Paper 1 & 2 15 Field Data collection 2 (Risk Predictor on CVD) 16 Data Analysis for Parameter Model 2,3,4 17 Develop MODEL 2 (Hospital Unit Cost Model) 18 Testing MODEL 2 19 Develop MODEL 3 (Pay-ability Model) 20 Testing MODEL 3 21 Develop MODEL 4 (Dynamic Clinical Pathway Model) 22 Testing MODEL 4 23 Analysis MODEL 2,3,4 & Report 3 24 Field Data collection 1b (Risk Predictor on CVD) 25 Tesis Writing & Mini Viva Preparation 26 Develop Proposal for get Ethical Approval 27 Obtain Ethical Approval 28 Field Data collection 2 (Hospital Unit Cost) 29 Field Data collection 3 (Pay-Ability) 30 Field Data collection 4 (Dynamic Clinical Pathway) 31 Data Analysis MODEL 2,3,4 & Tesis Writing / Report 4 32 Submit and Publish Paper 3 33 Develop Validation and Evaluation Mothod 34 Data collection and statical analysis 4 35 Analysis and evaluation 36 Tesis Writing & Viva Preparation

2017 9

10 11 12

2018 1

2

3

4

5

6

7

2019 8

9

10 11 12

1

2

3

4

5

6

7

8

9

10 11 12

Activities I. Study Literature (Learning) * Learning Health Financing Concept. * Learning DRG Concept - countries comparison. * Learning INA-CBGs Calculation System. * Learning Artificial Intelligent in Machine Learning. II. Disease Risk Predictor * Rules Algorithm for Risk Calculator # Stroke - Epidemiology study - Management Cost - Rule based Algorithm on Stroke Risk Calculator - Testing in few algorithm machine learning (Rapid Miner) # Cardiovascular Disease (CVD) - Epidemiology study - Management Cost - Rule based Algorithm on Stroke Risk Calculator - Testing in few algorithm machine learning (Rapid Miner and MATLAB) * Training and Prediction (simulation in MATLAB) # Build Stroke Risk Prediction # Build CVD Risk Prediction III. Unit Cost Calculation * Determine Fix Cost, Variable Cost, Semi-Fix Cost in Hospital * Determine Laboratory Testing Cost in Hospital * Determine Unit Cost # Stroke Management Cost (Validate or Recalculate) - Determine CP with Stroke Patient - Determine doctor Cost. # CVD Management Cost (Validate or Recalculate) - Determine CP with CVD Patient - Determine doctor Cost. IV. Pay-ability Calculation * Data personal Entry # Data Personal Entry and Validation # Patient Profiling * Simulate financial options for increasing pay ability # Determine Risk Prediction, then select Clinical Pathway or Healthy Pathway. # Calculate cost depend on Pathway selection. V. Clinical Pathway suggestion (Dynamic CP) * Create "Clinical Pathway and Health Pathway" Database # Stroke Case # CVD * Build Connection CP to Unit Cost Calculation VI. Build Medical Charge Prediction * Build Connection # Build Connection Risk Predictor, UC Calculation, Dynamic CP and Pay-Ability Calculation * Build Simulation for optimized cost # Build Rules-based Medical Charge Optimization # Simulate In MATLAB using Artificial Intelligent

Expected Result  This system is able to identify disease severity through disease risk classification.  This system is able to provide an alternative clinical pathway that is maximal from the available resources that are owned by the patient and the doctor will provide suggestions in their selection.  This system uses a Real Cost method and specific calculation system for each hospital so that cost calculations can be more accurate.  The system able to give alternative for health financing for patient, hospital and government.  This system is able to predict heath expenditure for any health stakeholder (patient, doctor, hospital and government).

Conclusion  The development of this system is expected to help government achieve the 100% UHC target.  This system is expected to solve the current problems, such as payment fairness, accuracy of disease classification and medical charge accuracy.  This system can provide health care cost prediction to decision makers to plan their health financing. • Government Health Expenditure Planning (for government). • Hospital Cost planning (for hospital). • Discharged Planning System (for patient).

Thank you