A Study on Singapore’s Ageing Population in the Context of Eldercare Initiatives Using Machine Learning Algorithms

Abstract

Ageing has always directly impacted the healthcare systems and, more specifically, the eldercare costs, as initiatives related to eldercare need to be addressed beyond the regular healthcare costs. This study aims to examine the general issues of eldercare in the Singapore context, as the population of the country is ageing rapidly. The main objective of the study is to examine the eldercare initiatives of the government and their likely impact on the ageing population. The methodology adopted in this study is Cross-Industry Standard Process for Data Mining (CRISP-DM). Reviews related to the impact of an ageing population on healthcare systems in the context of eldercare initiatives were studied. Analysis methods include correlation and machine learning algorithms, such as Decision Tree, Logistic Regression and Receiver Operating Characteristics curve analysis. Suggestions have been provided for various healthcare and eldercare systems’ initiatives and needs that are required to transform to cope with the ageing population.

1. Introduction

Though the country’s economic progress has been a success story, recently, Singapore is dealing with a rapidly ageing population, like many other developed countries around the world. The problem of an ageing population has been caused due to increasing longevity among older people and, at the same time, a decline in the birth rate [1]. The healthcare systems need to evolve to support these ageing populations, and various initiatives, such as eldercare, residential care, long-term care, acute care, etc., need to be advocated [2]. There needs to be a proactive approach from the various stakeholders to the impact of an ageing population on Singapore Healthcare Systems to arrive at necessary policy implications. Similarly, [3] underlined that Singapore is facing a significant challenge from the ageing population. There needs to be a variety of healthcare initiatives with policy changes across various domains with changing and challenging eldercare health needs and requirements. There is a crucial and urgent requirement to study the eldercare initiatives taken by the Singapore government in the context of an ageing population.

1.1. Population Growth and Structure in Singapore

It should be noted that Singapore is a small island country where the population in 1950 was 1,022,100, which kept growing to 1,646,400 in 1960 and 2,074,507 in 1970. In the past few decades, Singapore witnessed better economic growth; consequently, there was a significant rise in the population too. During 1980, 1990, 2000 and 2010, the population growth was, respectively, 2,413,945, 3,047,132, 4,027,887 and 5,076,732. From 1990, it can also be seen that the population figures have included the Permanent Resident Population and the Non-resident Population apart from the Singapore Citizen Population. The growing economy has made the government bring in more immigrant populations to ensure that the growth is not affected due to the limited number of Singaporeans.

It is clear from

Table 1. Population figures of Singapore: 2010 to 2019.

Variables	2010	2011	2012	2013	2014	2015	2016	2017	2018	2019
Total Population (Number)	5,076,732	5,183,688	5,312,437	5,399,162	5,469,724	5,535,002	5,607,283	5,612,253	5,638,676	5,703,569
Resident Population (Number)	3,771,721	3,789,251	3,818,205	3,844,751	3,870,739	3,902,690	3,933,559	3,965,796	3,994,283	4,026,209
Singapore Citizen Population (Number)	3,230,719	3,257,228	3,285,140	3,313,507	3,343,030	3,375,023	3,408,943	3,439,177	3,471,936	3,500,940
Permanent Resident Population (Number)	541,002	532,023	533,065	531,244	527,709	527,667	524,616	526,619	522,347	525,269
Non-resident Population (Number)	1,305,011	1,394,437	1,494,232	1,554,411	1,598,985	1,632,312	1,673,724	1,646,457	1,644,393	1,677,360
Total Population Growth (Per Cent)	1.8	2.1	2.5	1.6	1.3	1.2	1.3	0.1	0.5	1.2
Resident Population Growth (Per Cent)	1	0.5	0.8	0.7	0.7	0.8	0.8	0.8	0.7	0.8
Population Density (Total Population per square Kilometre)	7146	7273	7429	7540	7615	7697	7797	7796	7804	7866
Sex Ratio (Males per thousand Females)	974	972	970	968	967	965	963	961	959	957
Median Age of Resident Population (Years)	37.4	38	38.4	38.9	39.3	39.6	40	40.5	40.8	41.1
Median Age of Citizen Population (Years)	38.6	39.1	39.6	40	40.4	40.7	41	41.3	41.7	42
Natural Increase (Number)	20,357	21,627	24,182	20,782	22,839	22,323	21,234	18,710	17,757	na
Rate of Natural Increase (Per Thousand Residents)	4.9	5.1	5.6	4.6	5.1	4.9	4.6	4	3.7	na

Source: Department of Statistics Singapore.

that there has been a steady but marginal increase in the population from 2010 onwards. However, the rate of total population growth was 2.5% in 2012 and has subsequently declined since 2013. In 2017, it was 0.1%, and 2018 was 0.5%. The decreasing trend of the population growth rate was arrested with an increase in the rate to 1.2% in 2019. The resident population, including the Singapore Citizen Population, has been steady with a marginal increase over the last few years since 2012, though the Permanent Resident Population declined from 533,065 in 2012 to 525,269 in 2019. The non-resident population growth has also been steady, except for the decreasing trend in 2017 and 2018. It is also noted that the natural increase and the rate of natural increase have been showing a decreasing trend. Table 1 also shows that the Sex Ratio (Males per Thousand Females) had been decreasing from 974 in 2010 to 957 in 2019. The median age of the resident population has also shown an increasing trend as 2010 was 37.4 years, which grew steadily over the years, and in 2019 was 41.1. In a similar context, it can also be seen that the citizen population’s median age demonstrated an increasing trend from 2010 when the median age was 38.6 to 2019 when the median age was 42. This clearly shows that the population is ageing in Singapore, and there has been limited growth in the population of Singapore that would help in the changing demographic trend, which would be the ageing of the Singapore population. Similarly, the life expectancy rate went from 80.6 at birth to 82.6 at the age of 65. Additionally, the total fertility rate has remained below the replacement level at 1.2 births per female [4].

It is seen that when there is a decline in the fertility rate, there is an increase in the life expectancies. The country would experience some healthcare-related expenditure issues, and there is a possibility of affordability issues creeping up, as highlighted by [5]. A similar study conducted earlier by [6] indicated the necessity of greater access to healthcare and resources required for the ageing population in terms of healthcare cost. Another study conducted by [1] on The National Policy on Elderly People in Singapore examined that the expansion of longevity among the elderly and the decline in fertility contribute to an ageing population in Singapore, and Singapore’s elderly people are suffering from a high level of involuntary dependency which needs to be reviewed. It also stated that a consultative approach would help Singapore achieve higher standards of care for the increasingly wealthy elderly.

Problem Statement

Ageing has always directly impacted the healthcare systems and, more specifically, the eldercare costs, as initiatives related to eldercare need to be explicitly addressed beyond the regular healthcare costs. Such costs related to healthcare for the elder and its costs are significant. It is noted that Singapore had been spending about 11% each year from 2012 to 2017, and the expenditure related to healthcare has increased from $13 billion to $22 billion. Further, the operating healthcare costs also rise, and the workforce costs account for 60% of the costs. The need for eldercare and its associated costs is quite crucial, as it is noted that people aged above 65 years, on average, stayed 6.9 days in the public holidays, whereas the younger patients stayed for 3.9 days [7]. A study by [8] has also cautioned that there are several issues that the Singapore healthcare systems need to address for chronic and communicable diseases and ageing population issues. Even though Singapore’s healthcare systems have done well over the past 50 years and have been seen as one of the superior healthcare systems compared to the developed western countries, the research issue on eldercare initiatives and healthcare costs associated with eldercare needs to be studied for effective policymaking.

2. Review of Literature

It is important in a research study to review the current and available literature to understand the eldercare initiatives and the issues associated with them more clearly. An in-depth review of literature related to previously conducted research helped identify the research gaps, which the present study has focused on. Reviews related to the study are presented in the following section.

2.1. Healthcare and Eldercare Initiatives

Previous studies highlight that the policy initiatives related to healthcare, and particularly eldercare, need to be addressed in an ageing society. The authors of [9] highlighted that, usually, the healthcare of older people is generally under tremendous pressure due to rising healthcare costs; as such, the prevalent healthcare systems in terms of insurance are highly unfair. There is also evidence that the resources provided for eldercare are also insufficient as there are only a few hospitals and nursing homes for aged patients, leading to ineffective addressing of older people’s needs, which may go beyond non-communicable diseases and include disability and chronic illness for the elderly. Though the study is specific to China, the arguments stand relevant to many other countries and societies.

The author in the Ref. [10] agrees that the insurance related to eldercare needs to be addressed, as most societies’ support for eldercare is inadequate. Additionally, the study highlights how the Japanese government was able to initiate mandatory public long-term care insurance, which helped the older people to meet their healthcare needs, as the benefits provided do not differentiate people based on their income or financial situation. It is also extraordinarily liberal on both coverage and benefits, though there are challenges on home-based care, fiscal sustainability and relevant support for the family careers of older people. The study suggests that the rest of the world should use Japan’s expertise and strategies, as it argues that Japan’s eldercare policy is one of the most comprehensive long-term care systems for the world’s most ageing populations. A similar study conducted on the Nordic countries by [11] revealed that the healthcare reforms related to eldercare need to address the more significant healthcare needs. Most societies rarely evaluate the impact of equity in access to greater healthcare needs of older people.

Another study by [12] examined various provisions on the elderly care services with respect to Australia, which included a residential care model fully providing the geriatric health services. However, the problems in such eldercare are usually the coordination between various government bodies, including healthcare communities, and the services are typically fragmented. A study with reference to Taiwan conducted by [13] argues that elderly care services definitely assist in successful ageing, while [14] highlighted that the encouraging participation of older people in society needs to be part of the eldercare facilities.

2.2. Eldercare in Singapore

The authors of the Ref. [15] highlighted that by 2030, 19% of the Singapore population will be above 65 years, and there is an urgent need for a paradigm shift in regulating the eldercare sector in Singapore to cater to the changing demographics of the population. The study argued that the shift needs to be in the form of holistic, humanistic and multidimensional care for older people. The author of the Ref. [16] highlighted that the diseases in Singapore, which are chronic and non-communicable, such as coronary heart disease, stroke and cancer, would account for 60% of the death due to the rapidly ageing population and would present challenges to the healthcare delivery and financing systems in Singapore by 2020 and 2030.

Another study on the ageing population with respect to the future living arrangements of Singaporeans with age-related dementia by [17] agreed that the eldercare options for Singaporeans would need to ensure that it needs to accommodate the impact of the demographic shift on its healthcare and eldercare initiatives. Addressing the concerns proactively, the Ministry of Health in Singapore provides both intermediate services and long-term services in the form of residential healthcare services and community healthcare services. Residential healthcare services are provided in the community hospitals, chronic sick hospitals, nursing homes, inpatient hospice care and respite care. Community-based healthcare services are provided in both home hospice care services and also in the home medical and home nursing, which are home-based. Centre-based healthcare services provided within a centre include day rehabilitation centres, dementia daycare centres, psychiatric daycare centres and rehabilitation homes [18]. Further, the study by [19] has also indicated that the government is working to increase the number of nursing home beds to 14,000 by 2020, a 50% increase from 2010. A similar study by [20] reported that older adults with multiple chronic diseases are likely to demand more beds as the population grows older, and there is also a requirement for support with the ease of commuting and the proximity of service centres homes [21]. These studies indicate that there is a need to study the facilities for beds.

Evaluating various policies of the Ministry of Health, [22] studied the eldercare in Singapore with reference to how the country is using smart technologies to ensure that eldercare services are provided cost-effectively. The study concluded a need for a more collaborative and contextually sensitive approach to devise and develop smart eldercare solutions. In a similar context, another study conducted by [23] hoped that the government of Singapore, the various policymakers and the resource planners would review the health needs of the Singapore population proactively. The study identified that the likely demand for healthcare professionals and the various eldercare facilities are prevalent. The study has used the modelling approach and proposed eldercare requirements to address the strategic workforce planning in the healthcare sector. It provides projections required for the eldercare services among ophthalmic nurses, technicians and patient service assistants in Singapore.

As most of the households in Singapore need to work, older people depend on nursing homes to provide adequate custodial care, and the seniors are engaged in these homes. It increases the demand for nursing homes and is not only limited to nurses but also to other health professionals, including doctors [24]. The authors of the Ref. [25] indicated that due to unplanned hospitalization of older patients, there is a demand factor that needs to be addressed in the form of medical services, such as home medical care, doctor visits and hospitalization.

The study of [26] has a different view; it is indicated that eldercare policies should go beyond the healthcare needs of older people. Older people need to be encouraged for employment after retirement on either a full-time or part-time basis. They also integrate health and social services along with a reduction in the cost related to healthcare that has also been highlighted by [27] in their study on the Singapore eldercare facility in the context of over accommodation. A similar view was also reflected by [28], focusing on the interdependencies that would flow between the migration of care specialists in the area of costs and the possible issues of social change.

2.3. Research Gap

The review of the present literature indicates a fair amount of literature is related to the ageing of the population and its healthcare-related issues that the stakeholders need to address. However, there have been limited studies and differing perspectives on the healthcare needs of the elderly population. There have not been adequate studies from Singapore’s perspective, especially with the enormous scale of societal and demographic change that the country is undergoing. Further, the models adopted by Singapore have often been seen as the most tested and comprehensive initiatives, which most countries across the world replicate. Thus, a detailed and thorough study using big data tools is the need of the present situation.

2.4. Research Framework

The Cross-Industry Standard Process for Data Mining [29] has been used for the process of data mining exercise, which consists of the phases: business understanding, data understanding, data preparation, modelling, evaluation and deployment, and it is the framework of this study. From the current research problem perspective, which is the first step of the CRISP-DM (see Figure 1), the study has covered all the relevant improvements and steps in the healthcare systems in the context of the ageing population and also evaluated the initiatives related to eldercare, thereby ensuring that the objectives of the study have been appropriately addressed. In the second stage of CRISP-DM, the study has listed and collected the data sets required for the study. Data collection is not only limited to data loading but also understanding the data from multiple sources. Therefore, a holistic view of the research problems related to the study has been addressed, as it helps to see things in an integrated way rather than a disconnected and disjointed interpretation. Further, this leads to the third stage of CRISP-DM, where decisions related to data mining goals, availability of data, the volume of data, data types and the data attributes have been looked at, along with the quality of data and the constraints of obtaining such data. As the research choice used in the study is multi, in the fourth stage, the CRISP-DM looks at selecting the specific modelling technique for each task. The fifth stage of CRISP-DM applies the appropriate models, tools and applications. Moreover, it ensures that the research objectives are met in spite of the constraints on time and cost. It is noted that in the evaluation stage, the assessment of the results determined through data analysis is studied, whereas, in the development stage, various predictive analytics have been conducted to help achieve the objectives of the study.

Other models used in this study are regression, decision tree and random forest. Data mining has been used in this study as the gathered data has been analysed and produced, which helped process the data to sustain data analytics and predictive modelling [30].

2.5. Modelling

Common types of modelling in the data sciences for decision making are the analytics that relies on the nature of data, which is descriptive, predictive and prescriptive. The current research has used the data and worked on all three modelling techniques to examine and analyse the healthcare requirements of the ageing population. It is considered predictive rather than descriptive. This study has used machine learning algorithms, including Decision Tree, Logistic Regression and ROC (Receiver Operating Characteristics) curve.

The Decision Tree is a tree-structured classifier and is one of the supervised machine learning algorithms used for both classification problems and regression problems. The decision tree has an internal node that represents the dataset features, and the branches represent the decision rules. The outcome is represented with the leaf node. The decision tree algorithm is simpler as it is more like a human’s way of decision-making with less of a requirement of data cleaning than other machine algorithms. However, the process would be complex when the number of layers increases, leading to computational complexities. As such, the present study also uses other machine algorithms beyond the decision tree, which is Logistic Regression and Receiver Operating Characteristics (ROCC) curve analysis. Logistic Regression is a popular statistical model, which is a commonly and widely used classification algorithm in machine learning. Logistic regression uses binary classification, which translates the inputs to 0 or 1. 0 denotes negative class, and 1 denotes positive class. The input values (x) are usually referred to as Beta, which linearly combines using weights or coefficient values. The output values that are predicted as usually “y”. The equation for the Logistic Regression is as follows:

$$\mathrm{y}=\frac{{\mathrm{e}}^{{\mathrm{b}}_0+{\mathrm{b}}_1\times \mathrm{x}}}{1+{ \mathrm{e}}^{{\mathrm{b}}_0+{\mathrm{b}}_1\times \mathrm{x}}}$$

where y is the predicted output, b₀ is the bias or intercept term and b₁ is the coefficient for the single input value (x). Each column in the input data has an associated b coefficient (a constant real value) that must be learned from the training data.

Further, this research also used another machine learning algorithm, which is Receiver Operating Characteristics (ROC) curve analysis. ROC cure analysis uses a graphical plot to show the diagnostic ability of binary classifiers. The construction of the ROC curve is conducted by plotting the true-positive rate against the false-positive rate. It is a plot of the true-positive rate versus the false-positive rate at different classification cut-offs. [31]. As such, ROC curves are used when there are roughly equal numbers of instances for each class, in other words, when the data are balanced [32]. The present research uses Decision Tree, Logistic Regression and ROC (Receiver Operating Characteristics) curve to obtain a higher accuracy level in the results. Model building was used with the estimating model parameters. In this study, the training models in the machine learning algorithms, as well as data splitting and determining the parameters’ values, arrive at better results in the predictions. The process undertaken was to initially define the research problem with the objective to reach clarity and propose a solution, which was followed by constructing the data set, transforming the data, training a model, and then using the model to make the predictions.

3. Research Methodology

3.1. Research Questions

The ageing population issues are related to the careful policymaking on the various eldercare initiatives in the context of healthcare; the present study addresses the research questions provided below:

• How do the government’s eldercare initiatives likely impact the ageing population?
• What are the various healthcare systems initiatives and needs required to transform to cope with the ageing population?

3.2. Aims and Objectives

The study aims to examine the prevailing issues around eldercare in the Singapore context, as the ageing of the population is clearly seen. In this context and to address the research questions, the following objectives are presented in this study:

• To examine the eldercare initiatives of the government and their likely impact on the ageing population.
• To suggest various healthcare system initiatives and needs required to transform to cope with the ageing population.

Research methodology is an approach to systematically and methodically analytically solve the research problem. The present study’s research philosophy is Epistemology focusing on realism, where a quantitative method has been used to analyse the data and establish a relation to a scientific enquiry. A deductive approach has been used to arrive at the research questions, and the model has been used in the study after a literature review. The study is also inductive as it has developed a model, after gathering and then analysing the data to arrive at the predictive models using various data analysis and machine learning tools, which may be used in the healthcare and eldercare sectors in Singapore. The model may be looked at as a comprehensive theory if other countries see how healthcare and eldercare services are undertaken in Singapore. The research design used in this study is descriptive, as the study is based on existing secondary data, which are collected from various government sources. The justification for using secondary data can be underlined by the time and cost constraints, apart from the nature of the research area.

4. Results and Findings

4.1. Singapore Residents by Age Group

It is important for this study that the age group of the resident population needs to be studied, as the research focuses on the ageing population’s healthcare requirements. The ageing population is determined based on the number of residents in Singapore within the age segment that is growing older.

From

Table 2. Singapore Residents by Age Group: 2010 to 2019.

Age/Year	2010	2011	2012	2013	2014	2015	2016	2017	2018	2019
0–4 Years	194,432	188,249	186,673	183,295	181,369	183,575	187,160	187,653	185,528	185,355
5–9 Years	215,675	208,119	206,324	205,704	205,790	204,452	201,509	200,575	199,066	197,775
10–14 Years	244,302	240,536	233,320	226,244	220,796	214,388	207,495	206,253	206,530	207,926
15–19 Years	263,750	260,738	258,861	255,066	247,337	242,902	239,771	232,973	226,520	222,222
20–24 Years	247,190	255,069	264,506	266,889	263,918	264,127	260,854	259,072	255,516	249,079
25–29 Years	272,639	262,606	254,582	255,615	265,659	271,030	279,988	290,198	291,631	288,194
30–34 Years	298,687	297,555	295,396	296,870	292,287	290,619	285,544	279,340	281,397	293,446
35–39 Years	320,024	316,161	313,970	305,503	302,029	301,067	301,998	300,956	303,720	300,818
40–44 Years	309,441	306,614	308,823	312,581	316,023	316,755	313,445	311,484	303,640	300,763
45–49 Years	323,459	323,991	320,925	316,336	308,533	303,413	301,183	303,457	307,838	311,682
50–54 Years	303,044	308,130	310,376	312,988	315,503	315,091	315,598	312,814	308,965	301,682
55–59 Years	248,696	260,678	271,817	281,069	288,392	295,063	299,591	301,678	304,390	306,759
60–64 Years	191,995	208,160	213,996	222,154	231,502	240,493	251,853	262,651	271,688	278,828
65–69 Years	111,511	112,396	128,879	145,817	161,198	182,425	198,020	203,969	212,101	221,296
70–74 Years	92,618	100,073	104,414	105,746	106,815	102,631	103,796	119,494	135,546	150,127
75–79 Years	65,178	66,909	67,675	70,619	76,294	81,211	87,955	91,953	93,327	94,600
80–84 Years	39,839	42,145	44,115	46,199	48,418	51,785	53,556	54,341	57,103	62,293
85–89 Years	18,481	19,756	21,503	23,235	25,486	27,494	29,086	30,481	31,991	33,935
90 Years & Over	10,760	11,366	12,050	12,821	13,390	14,169	15,157	16,454	17,786	19,429

Source: Department of Statistics Singapore.

and Figure 2, it can be clearly understood that the age group of Singapore residents drastically changed from 2010 to 2019. In 2010, the differences between age groups had been more or less equally distributed among the groups. However, the subsequent years have witnessed a decline in the young population (see 0–4 years) and an increase in the middle-age population (see, 35–39 years, 40–44 years, 45–49 years, 50–54 years and 55–59 years). These age groups of Singapore residents contribute to more than 300,000 in each group, with the middle-age group of 45–49 years topping with 311,682.

4.2. Projection of Singapore Residents by Age Group

The analysis on the Singapore residents by age group from 2010 to 2019 indicates that the population is ageing, and it is important to see how the resident population will change over the next few years, which will help the policymakers frame the necessary action plan if required. Based on the data from 2010 to 2019, the population ages have been forecasted for 2020 to 2030. The base year set for comparison for the projection is 2019. To detect the seasonal pattern, FORECAST.ETS function in Excel 2016 is used. FORECAST.ETS function predicts a future value using Exponential triple Smoothing.

The projections as per

Table 3. Projection of Singapore Residents by Age Group: 2020 to 2030.

Age/Year	2020	2021	2022	2023	2024	2025	2026	2027	2028	2029	2030
0–4 Years	183,594	183,097	182,600	182,103	181,605	181,108	180,611	180,114	179,617	179,119	178,622
5–9 Years	195,594	193,975	192,356	190,737	189,118	187,499	185,880	184,261	182,642	181,023	179,404
10–14 Years	195,332	190,705	186,078	181,452	176,825	172,198	167,571	162,944	158,318	153,691	149,064
15–19 Years	218,579	213,773	208,967	204,161	199,354	194,548	189,742	184,935	180,129	175,323	170,516
20–24 Years	257,791	257,640	257,489	257,337	257,186	257,035	256,884	256,733	256,582	256,431	256,280
25–29 Years	293,206	296,840	300,475	304,110	307,745	311,379	315,014	318,649	322,284	325,919	329,553
30–34 Years	281,907	280,233	278,560	276,886	275,212	273,538	271,864	270,190	268,516	266,842	265,168
35–39 Years	295,408	293,369	291,330	289,290	287,251	285,212	283,172	281,133	279,094	277,054	275,015
40–44 Years	307,214	306,715	306,216	305,718	305,219	304,720	304,221	303,723	303,224	302,725	302,227
45–49 Years	300,182	298,019	295,855	293,692	291,528	289,365	287,201	285,038	282,874	280,710	278,547
50–54 Years	310,859	310,939	311,019	311,099	311,179	311,259	311,339	311,419	311,499	311,579	311,659
55–59 Years	320,483	326,787	333,090	339,394	345,697	352,001	358,305	364,608	370,912	377,215	383,519
60–64 Years	289,584	299,084	308,585	318,085	327,585	337,086	346,586	356,086	365,587	375,087	384,587
65–69 Years	242,404	255,975	269,547	283,118	296,690	310,261	323,833	337,404	350,976	364,547	378,118
70–74 Years	139,835	144,873	149,910	154,948	159,986	165,024	170,062	175,100	180,138	185,176	190,214
75–79 Years	100,507	104,313	108,119	111,926	115,732	119,538	123,345	127,151	130,957	134,763	138,570
80–84 Years	62,758	65,081	67,405	69,728	72,052	74,375	76,698	79,022	81,345	83,669	85,992
85–89 Years	35,784	37,537	39,289	41,042	42,795	44,547	46,300	48,053	49,805	51,558	53,310
90 Years & Over	19,430	20,356	21,282	22,208	23,134	24,060	24,986	25,912	26,837	27,763	28,689

Source: Department of Statistics Singapore.

and Figure 3 indicate that the population is ageing. In 2019, most of the population was between the age of 45–49 but was closely followed by the 50–54 and 55–59 age groups. It is also noted that the 35–39 years age group is not far behind. These age groups of people move towards the following age group segments, making the population grey. The replacement population in the younger age groups is also not healthy enough to lower the ageing population percentage. The analysis reveals that the 60–64 age group will be a clear majority of the population in 2030 and is very closely followed by the 55–59 and 65–69 age groups, making this research more relevant. The current situation’s needs enforce that adequate policy measures need to be implemented to cater to the ageing population’s requirements.

4.3. Comparison of Age Groups (65 Years and over versus Less Than 64 Years)

As it is clear from the Singapore residents’ age-wise analysis and from the projection till 2030 that the population is ageing, it is important to compare the population figures of the Singapore residents between 65 years and over versus less than 64 years. Figure 4 provides the comparison details of the above age groups, and

Table 4. Comparison of Age Group (65 Years and Over versus less than 64 years).

Age	2010	2011	2012	2013	2014	2015	2016	2017	2018	2019
65 Years & Over	338,387	352,645	378,636	404,437	431,601	459,715	487,570	516,692	547,854	581,680
Total less than 64 years	3,433,334	3,436,606	3,439,569	3,440,314	3,439,138	3,442,975	3,445,989	3,449,104	3,446,429	3,444,529
Total Residents	3,771,721	3,789,251	3,818,205	3,844,751	3,870,739	3,902,690	3,933,559	3,965,796	3,994,283	4,026,209
% age of residents 65 Years & Over	8.97%	9.31%	9.92%	10.52%	11.15%	11.78%	12.40%	13.03%	13.72%	14.45%

Source: Department of Statistics Singapore.

and

Table 5. Projection of Age Group (65 Years and Over versus less than 64 years).

Age	2020	2021	2022	2023	2024	2025	2026	2027	2028	2029	2030
65 Years & Over	600,718	628,136	655,553	682,971	710,388	737,806	765,223	792,641	820,058	847,476	874,893
Total less than 64 years	3,449,734	3,451,177	3,452,619	3,454,062	3,455,505	3,456,948	3,458,390	3,459,833	3,461,276	3,462,719	3,464,161
Total Residents	4,050,452	4,079,312	4,108,172	4,137,033	4,165,893	4,194,753	4,223,614	4,252,474	4,281,334	4,310,194	4,339,055
%age of residents 65 Years & Over	14.93%	15.54%	16.16%	16.78%	17.40%	18.02%	18.63%	19.25%	19.87%	20.49%	21.11%

Source: Department of Statistics Singapore.

provide the projection till 2030 calculated with 2019 as the base year.

Table 4 reveals that in 2019, the percentage of residents of the age group 65 years and over was 14.45%. The projection of the age group in Table 5 shows a gradual increase from 2020 to 2030. From the base year of 2019 (14.45%), the increase is seen as 15.54% in 2021, 16.78% in 2023, 18.02 in 2025, 19.25 in 2027 and 20.49% in 2029. Additionally, Figure 4 shows the comparison of age groups. It is also alarming to see the percentage moves drastically from 20.49% in 2029 to 21.11% in 2030.

4.4. Eldercare Facilities

When examining the eldercare initiatives, it is essential to look at the support that the old-aged people receive because older people are less economically active. As such, the working and economically active population would need to support the older aged people and economically inactive population. Further, the initiatives can be seen in the stakeholders spending on eldercare facilities; for instance, the data related to the beds for inpatient facilities would help.

4.5. Analysis of Old-Age Support Ratio

The examination of the old-age support ratio: residents aged 15–64 years per resident aged 65 years and over (per resident aged 65 years and over) is conducted in the study to understand the eldercare requirements and the initiatives that the policymakers need to consider, as support is key. The data are also presented with old-age support ratio: residents aged 20–64 years per resident aged 65 years and over (per resident aged 65 years and over), as the majority of the Singapore resident population becomes very much economically active from 20 years old onwards and thereby contribute to the national income and support for the residents 65 years and over, which is relevant and significant. The Ministry of Manpower in Singapore allows young persons aged 15–16 to work but with some restrictions with reference to the work in a non-industrial setting, and employers need to notify MOM along with the medical report submission requirements in case the work is in an industrial setting [33]. Hence, the data related to aged 15 and 20 are studied separately to see whether it makes any significant difference.

The projections in

Table 6. Old-age support ratio: residents aged 15–64 and 20–64 years per resident aged 65 years and over (per resident aged 65 years and over).

Ratios	2020	2021	2022	2023	2024	2025	2026	2027	2028	2029	2030
Old-age Support Ratio: Residents Aged 15–64 Years Per Resident Aged 65 Years and Over(Per Resident Aged 65 Years and Over)	4.4	4	3.6	3.3	2.9	2.5	2.1	1.8	1.4	1	0.8
Old-age Support Ratio: Residents Aged 20–64 Years Per Resident Aged 65 Years and Over(Per Resident Aged 65 Years and Over)	4.1	3.8	3.5	3.1	2.8	2.5	2.1	1.8	1.5	1.2	0.8

Source: Developed for the study.

reveal that the ratio continues to reduce, and as such, policymakers need to address the support for the old aged resident population with eldercare needs.

Table 7. Old-age support ratio: residents aged 15–64 and 20–64 years per resident aged 65 years and over (per resident aged 65 years and over).

Ratios	2010	2011	2012	2013	2014	2015	2016	2017	2018	2019
Old-age Support Ratio: Residents Aged 15–64 Years Per Resident Aged 65 Years and Over(Per Resident Aged 65 Years & Over)	8.2	7.9	7.4	7	6.6	6.2	5.8	5.5	5.2	4.9
Old-age Support Ratio: Residents Aged 20–64 Years Per Resident Aged 65 Years and Over(Per Resident Aged 65 Years & Over)	7.4	7.2	6.7	6.4	6	5.7	5.4	5.1	4.8	4.5

Source: Developed for the study.

and Figure 5 show that there has been a decrease in population support per resident aged 65 years and over from 7.4 (Aged 20–64) and 8.2 (Aged 15–64) in 2010 to 4.5 and 4.9, respectively. This shows an urgent requirement for eldercare initiative measures to be looked at, as the support group for the elderly population is shrinking.

4.6. Eldercare Facilities with Reference to the Resident Population from the Age of 65 Onwards

When examining the eldercare facilities in Singapore, the data related to the ‘Beds in Inpatient Facilities’ in the various types of hospitals in Singapore for the resident population aged 65 and over have been studied. The various types of eldercare facilities that have been examined are Acute Hospitals, Psychiatric Hospitals, Community Hospitals, Nursing Homes and Inpatient Hospices. The correlation of the resident population from the age of 65 onwards and the various eldercare facilities is provided in Figure 6. It highlights that the resident population aged 65 years and over has a positive correlation with the eldercare facilities in the context of ‘Beds in Inpatient Facilities’ in the Acute Hospitals, Community Hospitals, Nursing Homes and Inpatient Hospices, except for Psychiatric Hospitals. There is a negative correlation between psychiatric hospitals and eldercare facilities for residents aged 65 years (−0.910004).

The Heat Map in Figure 7 also shows positive correlations between the eldercare facilities in the context of ‘Beds in Inpatient Facilities’ in the Acute Hospitals, Community Hospitals, Nursing Homes and Inpatient Hospices, except for Psychiatric Hospitals with the resident population aged from 65 years. The cells in the GREEN colour indicate a positive correlation, while the PINK-coloured cells flag a negative correlation. The maximum negativity correlation is seen with reference to the Psychiatric Hospitals.

4.7. ROC Curve Analysis Using Decision Tree and Logistic Regression Method for Eldercare Facilities Focusing Beds in Inpatient Facilities

ROC stands for Receiver Operating Characteristics curve and has also been used to analyse the dataset related to eldercare in the area of ‘Beds in Inpatient Facilities’. The ROC curves are usually presented in the graph mode, and the area under the ROC curve indicates the value of using the analysis concerned.

Further, the Decision Tree and Logistic Regression method algorithms for analysis were tested using the ‘Beds in Inpatient Facilities’ data set. The output is provided in Figure 8.

The Decision Tree and Logistic Regression outputs are as follows:

roc_auc_score for Decision Tree: 0.817866718210672

roc_auc_score for Logistic Regression: 0.9229188841149525

Logistic Regression is superior to the Decision Tree method because the true positive rate is higher at all cut-offs, and the false positive rate is lower than for the Decision Tree method. The area under the curve for Logistic Regression is larger than the area under the curve for the Decision Tree method. Based on the result, the Logistic Regression method attained a good result compared to the Decision Tree ROC curve. Therefore, this indicates that beds in inpatient facilities need to be increased as the population will grow older, and they will be necessary to cater to the eldercare requirements of the ageing population.

5. Conclusions

The research clearly shows that there has been limited growth in the Singapore population that could help change the demographic trend, which is the ageing population of Singapore. The analysis with the base year as 2019 shows that the percentage of residents aged 65 years and over is 14.45%. The projection of the age group revealed that there is a gradual increase from 2020 to 2030, which policyholders need to deeply look at and react to. It can be seen that the eldercare facilities in the form of ‘Beds in Inpatient Facilities’ in Acute Hospitals, Community Hospitals, Nursing Homes and Inpatient Hospices have a positive correlation with the resident population aged 65 years onwards. This indicates that there needs to be continuous investment in such facilities as the population is ageing. Further, it can also be seen that there exists a negative correlation between eldercare facilities in the form of ‘Beds in Inpatient Facilities’ in Psychiatric Hospitals and the resident population aged 65 years and over. This may be because the awareness among the ageing population for such services and the investments in such facilities are not as high as the other healthcare facilities. The study also examined eldercare initiatives, and it is important to look at the support that old-aged people receive because as people become older, they become less economically active. The analysis on the old-age support ratio was performed, and further, the beds for inpatient facilities were also studied when examining eldercare initiatives. The results align with the “Silver Tsunami” projections presented by [34], which highlighted that, by 2030, nearly one in five (20%) adults would be aged 65 and older, and the percentage of US workers caring for an elderly person is expected to triple. Further, another study by [35] also supports the present study results, as it argues that caring for ageing parents while trying to manage demands from work and other life roles successfully will be crucial in the future, as the old-age support ratio will be challenging. The results also align with the ageing population issues in another South-East nation, Indonesia, which has increased from 4.5% in 1971 to 9.6% in 2019 and is thought to be a trigger for widespread community care for the elderly [36].

The methodology adopted in this study is CRISP_DM, where the data mining exercise is conducted in phases, namely business understanding, data understanding, data preparation, modelling, evaluation and deployment. The data plans required are data collection, data monitoring, data cleaning and filtering, data modelling, and deployment, leading to superior insights on the data, where data optimisation and data harmonisation are evident. Technology innovation using such big data is a continuous process that assists eldercare initiatives within Singapore and Community-Based Healthcare. The entire big data architecture would come together, with the Ministry of Health playing the regulatory role. Thereby, the model contributes to the accuracy of the results and, in turn, aids the decision making and assists the policymakers. Further, the study by Ageing International on eldercare in Asia [37] suggests that at least in Asian countries, ownership for support of working elder-caregivers is not yet determined. It is argued that government organizations, families, religious communities, social organizations, healthcare organizations and workplaces may bear some burdens in supporting employed caregivers. Similar to the present study, these researchers call for a more collaborative and integrative societal approach to supporting family-based informal eldercare in Asia and abroad.

5.1. Recommendations

Based on the results, findings and conclusions of this examination of the government’s eldercare initiatives, it likely impacts the ageing population, and we indicate the following recommendations. Firstly, increased spending is needed on the eldercare facilities in the ‘Beds in Inpatient Facilities’ in Acute Hospitals, Community Hospitals, Nursing Homes and Inpatient Hospices to meet the requirements of the resident population aged 65 years onwards. Secondly, with Logistic Regression showing good results compared to the Decision Tree ROC curve, there needs to be additional funding and resource allocations towards the Beds in Inpatient Facilities. This is crucial as such infrastructure needs to grow to cater to the eldercare requirements of the ageing population because the population will continue to grow older. Thirdly, the policymakers are also required to look at options for further investment in eldercare facilities for Psychiatric Hospitals, which need to be increased, as the current psychiatric facilities do not correlate with the requirements of the resident population aged 65 years onwards. Additionally, the awareness for such services related to the psychiatric requirements for the resident population aged 65 years onwards needs to be promoted. Targeted campaigns on the ageing population’s awareness and preparedness and the eldercare needs and facilities need to be widely undertaken. Lastly and most importantly, the Per-capita government health expenditure needs to be aligned, and the government needs to continuously spend to meet the healthcare requirements of the ageing population.

5.2. Limitations of the Study

The study used the data collected originally for different purposes and not exactly to understand the issues related to eldercare initiatives. As such, the relevance and reliability of the data in the context of the present study are debatable. Though consistency, relevancy and reliability of the data used cannot be assured due to its secondary nature, data sources have been verified for authenticity with data from only the relevant regulatory authorities used in this study. Despite the above limitation, there has been no compromise on the data mining models used in this study, and the research objectives of the study are appropriately addressed. Though stronger in the present study, data models would be even better if comprehensive big data architecture had been used to gain even more insights into the data.

5.3. Scope of Future Research

The study has examined the eldercare initiatives of the government, along with their likely impact on the ageing population; however, the study used 65 years old as a base year to evaluate the elderly initiative data. With the increase in life expectancy, future studies can examine the elderly initiatives and requirements for different categories, such as 75 years old and beyond, 90 years old and beyond, etc. Additionally, with communicable diseases, such as COVID-19, disrupting the healthcare costs and taking away much-needed attention from eldercare, studies related to such impacts can also be conducted. In addition, studies related to chronic illnesses that are likely to impact elderly individuals and the effects on the healthcare systems can also be studied.