Outdoor Space Elements in Urban Residential Areas in Shenzhen, China: Optimization Based on Health-Promoting Behaviours of Older People

Abstract

Given the ageing global population, it is important to promote “healthy ageing”. One of the most effective ways to achieve this is by prolonging the health of older people. Both the physical and mental well-being of older people are closely related to their living environment. Providing daily outdoor activities and enhancing the quality of public spaces and amenities in residential areas can encourage the adoption of healthy behaviours among older people. This study selected eight typical residential areas in Shenzhen, China, and analysed 40 outdoor public spaces. Video content obtained from fixed-point behavioural observation was entered into the Mangold INTERACT behavioural analysis system to extract the health behaviour data of older people. Regression analysis was then performed on the health behaviour data and the index data of the sample space elements. The results showed that several factors affect the outdoor health behaviours of older people. These factors include the scale of the outdoor space, the size of the hard ground area, the quality of the grey space, the green-looking rate, the accessibility of the site, the number of fitness facilities, and the richness of site functions. This study focused on a host of health-related behaviours such as rest, leisure, communication, and exercise. It confirmed the corresponding spatial needs of the elderly when engaging in the aforesaid activities. In this way, the quantitative research has supplemented previous studies by studying and evaluating the behaviour and activities of the elderly in specific settings. Through the analyses, a configuration model of outdoor space in residential areas was constructed with the aim of health promotion. Based on this model, a flexible and multilevel configuration list revealing seven specific types under three priorities is being proposed. The findings provide a scientific and effective strategy for optimising the quality of outdoor environments in residential areas. More specifically, the deployment of the Mangold INTERACT system to extract and quantify behavioural data enabled this study to overcome the limitations of traditional approaches to behavioural observation and recording. This provides a prelude for other quantitative research on the environment and behaviour.

1. Introduction

The World Health Organisation (WHO) predicts that by 2030, the number of people over 60 years old will increase to 1.4 billion, accounting for one-sixth of the world’s population. China faces a serious ageing problem with a rapidly growing elderly population ratio. In 2021, the percentage of the population over the age of 60 was 18.9%. It is estimated that by 2050, China’s elderly population will reach 487 million, accounting for 34.9% of the country’s population and one-quarter of the world’s elderly population [1]. To address the various challenges posed by population ageing, the WHO proposed “healthy ageing” in 2015. Japan, which was the first country in Asia to begin the population ageing process, realised as early as 1982, when it enacted the Health Care Law for the Elderly, that promoting health among older people and prolonging their health status as much as possible are the most effective means of achieving healthy ageing [2]. The main method of health promotion is health behaviour. “Health behaviour” refers to the behavioural measures taken by individuals to prevent diseases and maintain health, including various behaviours that actively promote health [3]. The health behaviour theory emphasises the individual as the anchor point and explores the synergistic effect of the interaction between the individual and the material and non-material environment on the development of individual health behaviours [4].

From the perspective of urban design, important means to improve the physical and mental health of older people include active intervention at the environmental level, actively guiding outdoor behaviours such as daily communication and exercise, and increasing the contact between older people and the outdoor natural environment. In both the United States and Japan, environmental intervention in “health promotion” behaviour is an important means of addressing ageing [5,6]. Active intervention in the environment needs to provide support for individuals’ health by inducing and supporting the occurrence of healthy behaviours [7,8]. Appropriate environmental design can provide residents with more opportunities to participate in outdoor activities, thereby directly or indirectly promoting emotional [9], physical, and mental health [10]. Green, a health researcher, clarified the potential environmental support for health promotion: “Health promotion is a combination of education and environmental support that promotes changes in behaviour and living conditions towards health” [11]. Many studies have confirmed this [12,13,14,15]. The encouraging or inhibitory effect of the environment on the healthy behaviour of individuals, especially older people, should not be overlooked. Scientific and rational environmental design or quality improvement strategies should be employed to promote the health status of older people.

In densely populated areas such as China, Singapore, South Korea, “Danchi” in Japan, and various metropolitan areas, congregate housing is the primary mode of living for urban residents. Outdoor public spaces in these residential areas are not only a transition between the residential unit and urban space but also an important place for older people to interact with nature and engage in healthy daily behaviours. The quality of these spaces can significantly influence the wellness of older people [16,17,18]. Similar to the development of urban construction in Western developed countries, China has transitioned from an era of mass building to a stage of stock building regeneration [19]. At the end of 2020, there were 49 billion square metres of stock housing in China, constituting a large proportion of residential areas. According to statistics [20], 52% of older people in China live in congregate housing. Due to their long construction histories, many residential areas can no longer meet the new needs brought about by rising living standards and emerging lifestyles. In particular, the outdoor space needs of older people with more leisure time cannot be met. Most outdoor spaces in these residential areas have a high vacancy rate, low vitality, and poor stayability, which seriously hinder the development of daily outdoor activities for older people. Clarifying the relationship between various health behaviours of older people and environmental elements and proposing strategies for allocating outdoor space elements in residential areas to promote healthy behaviours among older people is of great significance for establishing a liveable environment for older people in residential areas.

In the context of an ageing population, improving architectural spaces and environments is a long-term global challenge. Humanistic urban renewal requires systematic research on the physiological, psychological, and behavioural preferences of various user groups, particularly the elderly, disabled, children, and low-income individuals. These disadvantaged groups are often more sensitive to transformative changes in their environment [21,22,23]. Studies into the quantitative relationship between the environment and behaviour in outdoor public space can be broadly divided into two research directions. The first direction of research focuses on improving the overall vitality of public spaces. Such studies mostly examine the correlation between the number of visitors at a venue and environmental elements. In 2003, Rodiek surveyed the environmental preferences of elderly residents in welfare facilities during outdoor activities [24]. Sun Yi studied the quantitative relationship between the environmental characteristics of outdoor venues in residential areas and the number of active older people [25]. Gan Liang researched the quantitative relationship between diversity and vitality [26]. Wang Lijun investigated the quantitative relationship between the constituent elements of small and micro-urban public spaces and the frequency of spontaneous behaviours [27]. The second direction of research focuses on the correlation between specific behaviours and environmental elements. For example, Gavin R. McCormack and Alan Shiel studied the relationship between residents’ physical activity and the built environment of their community [28]. Yue Wenxiu conducted research on the correlation between residential site characteristics and the recreational and sports behaviour of older people [29]. Zhang Ran and Wang Hua researched the correlation between space and physical activity [30,31]. Wang He investigated the impact of residential areas and the surrounding built environment on residents’ interactive activities [32]. These studies have provided a wealth of empirical evidence for optimising the quality of outdoor public spaces. Through the analysis of mathematical statistical models, some researchers have identified the quantitative relationship between the behavioural activities of older people and factors such as path planning in space, the degree of greening, and the style and quality of surrounding buildings. Few studies, however, have explored differences in the degree of association between different types of behaviours among older people and spatial elements. For example, whether path planning that facilitates neighbourhood interaction behaviours among older people can also promote physical activity among older people has not been systematically explained and answered.

From a methodological perspective, research on the correlation between environment and behaviour is generally divided into two approaches: objective behaviour observation [25,33,34,35] and subjective preference surveys [29,32,36]. Objective behaviour observation methods typically employ techniques such as behaviour observation, behaviour annotation, and behaviour clustering to collect and extract data. While these traditional research methods have been used for a long time, they can be difficult to scale when dealing with large samples. This also makes it challenging for behavioural research to analyse data from a multidimensional perspective, with most studies limited to discussing a single target variable. For example, generalised variables such as vitality, number of visitors, behaviour frequency, and behaviour diversity can obscure the inherent differences in individual needs for spatial and morphological characteristics for different behavioural purposes. This can result in a lack of in-depth exploration of the complex combination of environmental elements in the context of multidimensional behaviour. Environmental surveys based on individual subjective preferences typically use methods such as questionnaire surveys, structured or semi-structured interviews, and picture preference ranking to collect and extract behavioural preference data. The advantage of these methods is that they can overcome the limitations of existing conditions and broaden the scope of exploration. However, the research may also be affected by the credibility and validity of the questionnaire. Thus, among studies on improving the environmental quality of urban residential areas, there is a lack of systematic empirical investigations aimed at promoting the health of older people. Additionally, environmental investigation methods based on individual objective behaviours also need to be improved.

Based on the empirical research framework above, this study used the Mangold INTERACT behavioural analysis system to collect behavioural data. The INTERACT system uses video analysis software for behaviour observation. With its built-in coding and behavioural timeline functions, the INTERACT system is equipped to quantify the quality of multicategory behaviours in the same scene. The data generated by the system can indicate the precise timing and process of behaviour. In previous studies, the behaviour analysis system has been shown to be of great value in the field of behaviour observation analysis in psychology, social empirical research, and cognitive neuroscience [37,38,39]. It can efficiently sample and analyse all behaviours that occur in large sample scenarios and can recode them into data variables that explain the frequency and duration of behaviours according to research needs. This is difficult for traditional behavioural investigation methods to achieve.

For its sample, this study used Shenzhen, a city that is representative of China’s economic development. Forty outdoor public spaces in eight typical residential areas were selected as samples, and various behavioural data on older people were collected and extracted with the help of Mangold INTERACT. A quantitative model was established between spatial elements and the health behaviours of older people through mathematical statistical model calculations, and the potential mechanism of action between multidimensional behaviours and residential spatial elements was explored. This paper also proposes a spatial design strategy to promote healthy behaviours among older people, and a flexible and adjustable environmental element configuration list was constructed to provide a scientific and precise optimisation strategy for the design of outdoor public spaces. The research method can also serve as a reference for behavioural data extraction and quantification methods in other environmental quality optimisation studies.

The paper presents the research development of this case from the Materials and Methods, Results, Discussions, and Conclusions. The Materials and Methods section describes in detail the research methods and framework, a sample overview, and specific definitions of the health behaviours of older people and environmental elements. In the Results section, the paper demonstrates the process of acquiring, processing, and computationally modelling behavioural and environmental indicator data, including the workflow of Mangold INTERACT in the data extraction process. The Discussion section summarises and analyses the results, identifies the specific needs of four types of health behaviours with regard to spatial elements, and proposes a list of spatial element configurations that can effectively promote healthy behaviours among older people based on these findings. This section also includes a summary of the strengths and limitations of the study. Finally, the Conclusion reviews and summarises the work and discusses the potential value and contribution of this research to international research fields.

2. Materials and Methods

2.1. Research Methodology and Framework

The purpose of this study is to understand how outdoor space elements in urban residential areas influence the behaviour of older people. An environmental optimisation strategy guided by health promotion is proposed. The research content and steps are as follows (Figure 1):

(I)

Forty outdoor public space samples with typical environmental characteristics were selected in stock residential areas in Shenzhen, China;

(II)

Continuous video recording of these 40 samples was conducted during the same time period to record the occurrence of various behavioural activities in older people. The videos were then imported into the Mangold INTERACT system to extract health behaviour data according to the analysis requirements;

(III)

A quantitative analysis was carried out on previous research. Additionally, the morphological elements and index quantification methods of outdoor space in residential areas that can be perceived by older people were summarized. According to specific quantification methods, data extraction of various element indicators was carried out for each of the 40 sample scenes;

(IV)

The two types of data, element and behaviour, were fitted to an ordered logistic regression equation in SPSS. This allowed for the calculation and establishment of a relationship model between the configuration of spatial elements and the health behaviours of older people. The environmental conditions that were conducive to the development of various health behaviours among older people were then clarified.

2.2. Selection of Scene Sample

A sample of 40 outdoor public spaces in eight established settlements in Shenzhen is used to investigate the influence of outdoor space elements on the activities of older people (Figure 2 and Figure 3). In order to ensure that the samples have the typical environmental characteristics of existing residential areas in China, the selection of samples is based on the following four principles (

Table 1. Sample selection information.

Residential Area	1. Yugui	2. Lanyuan	3. Zizhu	4. Zhaoyin	5. Sihai	6. Liyuan	7. Xuefu	8. Yitian
Location	Nanshan	Nanshan	Nanshan	Nanshan	Nanshan	Nanshan	Nanshan	Futian
Built Time	1994	1995	1989	1994–1998	1986–1998	1996	1999	1994
FSR	2.50	2.60	2.50	2.90	2.20	2.00	3.10	3.00
Aging Rate	9.8%	8.2%	10.2%	9.6%	8.2%	8.5%	8.1%	12.6%
Sample Size	7	2	4	2	3	6	3	13
Total	40

): (i) Location and distribution of the residential areas: Nanshan District and Futian District in the centre of Shenzhen were selected as research objects. The characteristics of stock residential areas in these areas are similar, with typical features such as early construction, high degree of completion, convenient location but single planning function, scattered spatial layout, and insufficient environmental sharing. (ii) Completion time of the residential areas: Based on the definition of existing residential areas in the “Guiding Opinions of the General Office of the State Council on Comprehensively Promoting the Reconstruction of Old Urban Residential Areas”, the construction time of samples was controlled between 1980 and 2000. (iii) The floor area ratio of the residential areas: Referring to the construction standards of multi-story residential areas in China, the floor area ratio of the residential area where samples are located was controlled between 2.0 and 3.2. (iv) Ageing rate of the residential areas: Combined with the ageing rate of Shenzhen, the ageing rate of the residential area where scenes are located was controlled between 8% and 13%.

2.3. Healthy Behaviour of Older People

During the period from 1 March to 20 April 2023 (22–26 °C), between 9:00–11:00 and 15:30–17:30 (avoiding mealtime), the research team conducted continuous video sampling of various behavioural activities of the older people in 40 samples. The recording time of each sample was not less than 2 consecutive hours. Through fixed-location video shooting, the real daily lives of the older people were clearly recorded. The collected videos were then input into the Mangold INTERACT behavioural analysis software for event coding, recording, and data visualisation analysis (Figure 4).

Health behaviour theory applies to a wide range of activities, such as physical activity, environmental protection, diet control, personal care, smoking cessation, disease self-examination, and safe sex, which together form healthy lifestyles [40]. Through field research, it was found that the behaviours of older people in outdoor public spaces in residential areas include not only personal behaviours such as walking, sunbathing, and meditation but also social behaviours such as board games, sitting and chatting, and square dancing. Different behavioural motivations determine the differences in individual needs for spatial morphological arrangement. To study the correlation between health behaviours and environmental composition, it is useful to determine what behaviours improve health. Using Zhao’s classification of the outdoor daily activities of older people [41], this study divides the behavioural motivations generated by the outdoor public spaces into the following seven categories: (i) Behavioural activities for the purpose of handling personal or family daily affairs, such as waste removal and carrying goods; (ii) behavioural activities for the purpose of rest and recuperation, such as meditation, sunbathing, and napping; (iii) behavioural activities for leisure and recreational purposes, such as playing chess, playing cards, singing, and walking the dog; (iv) behavioural activities for the purpose of social interaction, such as outdoor gatherings and sitting and chatting; (v) behavioural activities for the purpose of exercise and recreation, such as doing exercises, square dancing, and badminton; (vi) behavioural activities for the purpose of childcare, such as taking children to play in the sand, to play on swings, and to play games; and (vii) behavioural activities for the purpose of community work, such as workshops and landscaping. According to the definition of healthy behaviours [42], rest, leisure, communication, and exercise are health-promoting behaviours for older people. This study focuses on these four types of behaviours, and the specific activities are categorised in Figure 5 and

Table 2. Coding of healthy behaviours and interpretation.

Type	Motivation	Content
1. Rest Behaviour	Rest and recuperation	Sunbathing, napping, meditation, etc.
2. Leisure Behaviour	Entertainment and recreation	Playing chess, mahjong, poker, reading, dog walking, etc.
3. Communication Behaviour	Social interaction	Outdoor gatherings, chatting, tea breaks, etc.
4. Exercise Behaviour	Physical fitness	Square dance, Tai Chi, running, equipment exercise, badminton, table tennis, etc.

. Health behaviour data in the sample were extracted and quantified by recoding in the INTERACT behaviour analysis system.

2.4. Configuration of Elements in Outdoor Public Spaces of Residential Areas

Environmental supports that improve health are often formed by a combination of multiple types of physical elements. Given the sample size, it was not possible to examine all the variables of the environmental design attributes. Instead, 11 components of outdoor space that can be perceived by older people were selected. The selection criteria were based on the literature. The study used the Web of Science and CNKI databases, and advanced searches were conducted using themes and keywords such as “elderly + residential environment”, “elderly + outdoor public space”, and “elderly + residential green space”. Thirty studies related to improving the quality of residential public spaces from the perspective of older people were selected, and the physical environmental factors included in these studies were quantitatively analysed. The 12 environmental elements with the highest frequency were extracted (

Table 3. Frequency statistics of related research on environmental elements.

Environment Elements	Literature Reference Sources	Frequency
1. Recreational Facility	[27,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57]	16
2. Green Looking Rate	[43,44,45,47,49,51,53,55,57,58,59,60,61,62,63]	15
3. Percentage of Ground Pavement	[25,26,50,52,54,55,56,57,62,63,64,65,66,67]	14
4. Fitness Facility	[25,26,46,50,54,62,63,66,67]	9
5. Functional Diversity	[47,51,52,54,56,58,61,66,68]	9
6. Percentage of Pavilion Coverage	[25,47,51,53,55,57,65,68]	8
7. Plant Diversity	[26,51,52,53,56,58,63,65]	8
8. Lighting Facility	[50,51,52,53,56,63,66,67]	8
9. Area	[25,51,52,55,56,57,62]	7
10. Spatial Connectivity	[47,49,51,55,60,62]	7
11. Colour Uniformity	[26,43,53,62,63,67]	6
12. Spatial Enclosure	[25,46,56,59,64]	5

). Since this study was conducted during the daytime, “lighting facilities” were excluded. The final selection of 11 environmental variables was as follows: “Recreational facilities”, “Green looking rate”, “Percentage of ground pavement”, “Fitness facilities”, “Functional diversity”, “Percentage of pavilion coverage”, “Plants diversity”, “Site area”, “Spatial connectivity”, “Colour uniformity”, and “Degree of spatial enclosure”. These variables are classified as space composition, landscape composition, or facility composition (Figure 6).

3. Results

3.1. Behavioural Data Quantification

Quantifying the occurrence of healthy behaviours in older people involves combining the two dimensions of occurrence frequency and duration [69] to define the behaviour of a single research subject. Taking the occurrence of rest behaviour Y1 as an example, Y1 is equal to the behaviours of basking in the sun, closing eyes, napping, and meditation behaviours. The higher the occurrence of behaviour based on a certain health motivation, the more inclined the environmental conditions are to support the health benefits of older people through this type of activity. The sum of the four types of behaviours—rest, leisure, communication, and exercise—is the total number of health promotion behaviours that occurred in this sample scenario. This indicator explains the degree of environmental support for the health promotion of older people (Figure 7). The ratio of the amount of each subtype of behaviour to the total amount of health promotion behaviour reflects the degree to which the scene supports the occurrence of this type of behaviour. The data extraction and calculation processes were completed on the Mangold INTERACT behavioural analysis platform (Figure 8). The calculation is as follows:

$$\begin{gathered} \mathrm{Y}={\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{p}}_{\mathrm{i}}{\mathrm{t}}_{\mathrm{i}} \\ \mathrm{S}={\sum }_{\mathrm{i}=1}^4{\mathrm{Y}}_{\mathrm{i}} \\ \mathrm{R}=\frac{\mathrm{Y}}{\mathrm{S}}\ast 100\% \end{gathered}$$

where Y is the occurrence of various behaviours, P is the sample of the older people per unit, t is the duration of the behaviour of the older people per unit, S is the total number of health behaviours in the sample of the unit scene, and r is the ratio of various types of behaviours to the total number of health behaviours.

3.2. Quantification of Spatial Elements

In this study, the specific quantification methods for the 11 types of spatial elements refer to the previous literature, and the index data mainly came from on-site surveys and measurements. The plant diversity, the functional diversity, and the number of recreational and fitness facilities were measured by on-site or photo counting. Indicators such as area, the percentage of ground pavement, the percentage of pavilion coverage, spatial connectivity, and enclosure were collected with the help of rangefinders and drones. The green looking rate and colour uniformity were obtained by extracting the proportion of pixels from a professional analysis platform and substituting them into relevant algorithms. The specific measurement and quantification methods are shown in

Table 4. Quantification methods for environmental elements.

Type	Environment Elements	Quantification Method	Data Sources	Reference
Space Composition	1. Area	Total area occupied by the sample space in square meters.	On-site measurement	[70]
	2. Percentage of Ground Pavement	Ratio of the pavement area of building materials such as concrete, ceramic tiles, floor tiles, cast-in-place grass bricks, and embossed floors to the total floor area of the sample space.	On-site measurement	[71]
	3. Percentage of Pavilion Coverage	Ratio of the vertical projected area of the pavilion covering to the total area of the sample space.	On-site measurement	[27]
	4. Spatial Connectivity	Number of other outdoor public spaces in the residential area that are directly connected to the sample space.	On-site measurement	[48]
	5. Spatial Enclosure	Ratio of the sum of the sides surrounded by buildings to the total side length of the sample space.	On-site measurement	[72]
Landscape Composition	6. Green looking Rate	Average green looking rate in the four fields of view.	Sampling calculation	[73]
	7. Plant Diversity	Plant diversity R was calculated using the Patrick index: R = S, where S represents the number of plant species in the sample space. Due to the difference in the area of the sample space, the formula was adjusted to R = H/lgA, where A is the total area of the sample space.	On-site measurement	[74]
	8. Colour Uniformity	The Simpson index was used to calculate the colour uniformity. The image in the sample space was converted into a pixel map, and the proportion of colour pixels in the image was extracted. The calculation is ${1-\sum _{\mathrm{i}=1}\left(\frac{{\mathrm{q}}_{\mathrm{i}}}{{\sum }_{\mathrm{i}=1}^{\mathrm{I}}{\mathrm{q}}_{\mathrm{i}}}\right)}^2$, where ${\mathrm{q}}_{\mathrm{i}}$ is the total pixels of elements of type i, and I is the total pixels of the visible range.	Sampling calculation	[75]
Facility Composition	9. Functional Diversity	The functional diversity F was calculated using the Patrick index: F = N, where N represents the number of facility types in the sample space. Due to the difference in the sample space area, the formula was adjusted to F = N/lgA, where A is the total area of the sample space.	On-site measurement	[75]
	10. Recreational Facilities	Leisure facilities include wooden chairs, stone benches, and flower beds that support people to sit and rest. In the sample scene, these facilities are typically arranged in group settings and thus this indicator is calculated in units of “groups”.	On-site measurement	[43]
	11. Fitness Facilities	Fitness facilities include exercise equipment such as running machine, treadmills, and table tennis tables. In the sample scene, these facilities are typically arranged in group settings and thus this indicator is calculated in units of “groups”.	On-site measurement	[76]

. Combining the various methods above, a total of 11 types of environmental element index data were obtained in 40 samples. Table 4 shows the specific quantification methods.

3.3. Data Preprocessing and Verification

The data for the 11 spatial elements and the behaviour data of the population of older people in the 40 sample spaces were entered into the data analysis software SPSS. The generalised ordered logistics model was selected to analyse the explanatory effect of the 11 environmental independent variables on the behaviour variables (target variables). Before formal analysis, the raw data were processed in two steps:

(I)

The behavioural variables (dependent variables) with precise numerical raw data were transformed from continuous variables into hierarchically ordered categorical variables. The total amount of health promotion behaviours is divided into data cut points at equal intervals of 33.33%, and the three grades of low, medium, and high are assigned according to the range of values; the proportion data of rest, leisure, communication, and exercise activities is divided into data ranges based on the actual data distribution characteristics. The scope sets the dividing points: less than or equal to 20% as “unfavourable behaviour”, greater than 20% and less than or equal to 40% as “secondary behaviour”, and greater than 40% as “primary behaviour” (Figure 9);

(II)

The original data of various environmental element variables (independent variables) were standardised by Z-score.

After the pre-processing of the original data, the independent variable and the dependent variable met the input conditions of the generalised ordered logistics model. Before the regression analysis, however, it was necessary to ensure that there was no multicollinearity among the independent variables. In the multicollinearity test results, the tolerances of all independent variables were >0.1, and the variance inflation factor (VIF) was <5. Thus, there was no multicollinearity among the independent variables, and the ordered logistic regression analysis could be performed (

Table A1. Collinearity statistics of independent variables.

Model		Tolerance	VIF
	Area	0.590	1.696
	Percentage of Ground Pavement	0.490	2.040
	Percentage of Pavilion Coverage	0.768	1.303
	Spatial Enclosure	0.420	2.383
	Spatial Connectivity	0.411	2.435
	Green looking Rate	0.432	2.317
	Plant Diversity	0.506	1.976
	Color Uniformity	0.691	1.447
	Recreational Facilities	0.418	2.395
	Fitness Facilities	0.642	1.559
	Functional Diversity	0.547	1.829
a. Dependent Variable: S

).

3.4. Regression Analysis for the Total Amount of Health Promotion Behaviour

In the regression analysis of the total amount of health promotion behaviour, the total amount of health behaviour (S) and the 11 environmental elements were used as dependent variables and independent variables in the regression model. The Omnibus test passed (p ≤ 0.05), which means that at least one model parameter was significant. There are factors that can explain the difference in behaviour among the 11 environmental elements (

Table A2. Omnibus test.

Likelihood Ratio Chi-Square	df	Sig.
48.074	11	0.000

).

The results show that the total amount of health promotion behaviours (S) of older people in the public space is significantly correlated with the area, spatial connectivity, green looking rate, and default feature diversity (p ≤ 0.05), as shown in

Table 5. Ordinal logistic regression analysis with OR values and 95% Wald confidence intervals: healthy behaviour.

Parameter				95% Wald Confidence Interval for Exp(B)
		Sig.	OR(Exp(B))	Lower	Upper
Threshold	[S = low]	0.017	0.131	0.025	0.700
	[S = medium]	0.001	23.639	3.375	165.578
Area		0.002 *	23.779	3.138	180.173
Percentage of Ground Pavement		0.825	0.872	0.259	2.934
Percentage of Pavilion Coverage		0.302	1.633	0.644	4.143
Degree of Spatial Enclosure		0.739	1.251	0.335	4.666
Spatial Connectivity		0.012 *	7.328	1.548	34.697
Green Looking Rate		0.032 *	0.143	0.024	0.842
Plant Diversity		0.396	1.857	0.444	7.761
Color Uniformity		0.656	1.294	0.416	4.021
Recreational Facilities		0.524	0.675	0.202	2.261
Fitness Facilities		0.239	0.510	0.166	1.567
Functional Diversity		0.004 *	18.239	2.535	131.229

* The parameter is significant at the level of 0.05.

. It is positively correlated with the area (OR: 23.779 > 1), space connectivity (OR: 7.328 > 1), and functional diversity (OR: 18.239 > 1) and negatively correlated with the green looking rate (OR: 0.143 < 1).

3.5. Regression Analysis of the Proportions of the Four Subcategories of Behaviours

In the regression analysis of the proportions of the four subcategories (rest, leisure, communication, and exercise), the significance of the Omnibus test was less than 0.05, which means that at least one of the model parameters was significant. There are factors that can explain the differences in the proportions of the four behaviours among the 11 environmental elements.

(1) The proportion of rest behaviours (r1) was significantly correlated with the area, colour uniformity, and number of fitness facilities (p ≤ 0.05), as shown in

Table 6. Ordinal logistic regression analysis with OR values and 95% Wald confidence intervals: rest behaviour.

Parameter				95% Wald Confidence Interval for Exp(B)
		Sig.	OR(Exp(B))	Lower	Upper
Threshold	[r1 = unfavorable]	0.906	1.069	0.351	3.257
	[r1 = secondary]	0.000	63.104	8.567	464.838
Area		0.033 *	3.242	1.098	9.575
Percentage of Ground Pavement		0.471	0.672	0.228	1.980
Percentage of Pavilion Coverage		0.084	0.216	0.038	1.225
Degree of Spatial Enclosure		0.620	1.357	0.407	4.523
Spatial Connectivity		0.083	0.328	0.093	1.158
Green Looking Rate		0.577	0.715	0.220	2.323
Plant Diversity		0.352	1.791	0.525	6.110
Color Uniformity		0.034 *	0.324	0.114	0.918
Recreational Facilities		0.425	1.626	0.493	5.366
Fitness Facilities		0.023 *	0.148	0.029	0.769
Functional Diversity		0.354	0.605	0.209	1.749

* The parameter is significant at the level of 0.05

. It is positively correlated with the area (OR: 3.242 > 1) and negatively correlated with colour uniformity (OR: 0.324 < 1) and the number of fitness facilities (OR: 0.148 < 1).

(2) The proportion of leisure behaviour (r2) was significantly correlated with the percentage of ground pavement, pavilion coverage, and functional diversity (p ≤ 0.05), as shown in

Table 7. Ordinal logistic regression analysis with OR values and 95% Wald confidence intervals: leisure behaviour.

Parameter				95% Wald Confidence Interval for Exp(B)
		Sig.	OR(Exp(B))	Lower	Upper
Threshold	[r2 = unfavorable]	0.002	0.217	0.084	0.563
	[r2 = secondary]	0.000	15.203	4.021	57.479
Area		0.869	0.925	0.368	2.325
Percentage of Ground Pavement		0.027 *	3.433	1.151	10.243
Percentage of Pavilion Coverage		0.033 *	3.161	1.095	9.125
Degree of Spatial Enclosure		0.148	2.345	0.740	7.436
Spatial Connectivity		0.699	0.802	0.262	2.453
Green Looking Rate		0.618	1.320	0.443	3.932
Plant Diversity		0.625	0.781	0.290	2.106
Color Uniformity		0.053	2.462	0.989	6.131
Recreational Facilities		0.697	1.244	0.415	3.731
Fitness Facilities		0.667	1.215	0.501	2.942
Functional Diversity		0.038 *	0.301	0.097	0.936

* The parameter is significant at the level of 0.05

. It is positively correlated with the proportion of ground pavement (OR: 3.433 > 1) and pavilion coverage (OR: 3.161 > 1) and negatively correlated with functional diversity (OR: 0.301 < 1).

(3) The proportion of communication behaviours (r3) was significantly correlated with the number and functional diversity of fitness facilities (p ≤ 0.05), as shown in

Table 8. Ordinal logistic regression analysis with OR values and 95% Wald confidence intervals: communication behaviour.

Parameter				95% Wald Confidence Interval for Exp(B)
		Sig.	OR(Exp(B))	Lower	Upper
Threshold	[r3 = unfavorable]	0.001	0.162	0.055	0.478
	[r3 = secondary]	0.001	7.338	2.348	22.932
Area		0.823	1.105	0.459	2.662
Percentage of Ground Pavement		0.439	0.646	0.214	1.954
Percentage of Pavilion Coverage		0.580	0.762	0.292	1.993
Degree of Spatial Enclosure		0.717	1.237	0.392	3.901
Spatial Connectivity		0.332	1.780	0.555	5.707
Green Looking Rate		0.518	1.440	0.477	4.350
Plant Diversity		0.835	0.901	0.340	2.393
Color Uniformity		0.053	2.865	0.987	8.320
Recreational Facilities		0.657	1.324	0.383	4.585
Fitness Facilities		0.004 *	0.115	0.026	0.495
Functional Diversity		0.002 *	42.256	3.858	462.846

* The parameter is significant at the level of 0.05

. It is positively correlated with functional diversity (OR: 42.256 > 1) and negatively correlated with the number of fitness facilities (OR: 0.115 < 1).

(4) The proportion of exercise activities (r4) was significantly correlated with space connectivity, green-looking rate, plant diversity, and the number of fitness facilities (p ≤ 0.05), as shown in

Table 9. Ordinal logistic regression analysis with OR values and 95% Wald confidence intervals: exercise behaviour.

Parameter				95% Wald Confidence Interval for Exp(B)
		Sig.	Exp(B)	Lower	Upper
Threshold	[r4 = unfavorable]	0.234	0.546	0.202	1.480
	[r4 = secondary]	0.002	7.253	2.109	24.948
Area		0.559	1.362	0.483	3.838
Percentage of Ground Pavement		0.884	1.090	0.342	3.473
Percentage of Pavilion Coverage		0.811	1.102	0.495	2.453
Spatial Enclosure		0.583	0.714	0.215	2.374
Degree of Spatial Connectivity		0.015 *	4.941	1.357	17.998
Green Looking Rate		0.015 *	0.134	0.026	0.681
Plant Diversity		0.038 *	5.287	1.095	25.528
Color Uniformity		0.113	0.395	0.125	1.245
Recreational Facilities		0.228	0.474	0.141	1.598
Fitness Facilities		0.019 *	6.246	1.359	28.705
Functional Diversity		0.781	0.852	0.275	2.635

* The parameter is significant at the level of 0.05

. It was positively correlated with space connectivity (OR: 4.941 > 1), plant diversity (OR: 5.287 > 1), and the number of fitness facilities (OR: 6.246 > 1) and negatively correlated with the green-looking rate (OR: 0.134 < 1).

4. Discussion

Participating in activities in the outdoor public spaces of residential areas are one of the most important ways for older people in urban areas to improve their health. In this regard, the composition of the environmental elements is particularly important. From the analysis results, the environmental elements that attract older people to come into contact with nature and participate in health-promoting activities are (i) spatial scale, (ii) the scope of hard pavement on the ground, (iii) grey space quality, (iv) green looking rate, (v) accessibility of site space, (vi) the amount of fitness equipment, and (vii) the richness of the site’s preset functions. It was found that the natural landscape in the outdoor space of the residential area cannot be the only factor that determines the environmental health benefits. This finding is different from previous studies [15,77]. Moreover, the design of outdoor public spaces to improve health should consider the configuration of elements and match them according to the people’s needs.

4.1. Behavioural Motivation and Configuration of Spatial Elements

This study found that older individuals actively select locations with compatible spatial elements to engage in activities that align with their specific behavioural needs. Furthermore, the compatibility of spatial elements varies according to differing behavioural requirements. For example, the study found a significant relationship between the motivation for rest behaviour and the configuration of spatial elements, such as the size of the site area, colour uniformity, and the number of fitness facilities. These findings are consistent with previous research [41]. Older people tend to engage in healthy behaviours for rest and recuperation in spaces with a relatively large spatial scale and a visual focus. However, an excessive amount of fitness equipment does not support the occurrence of such behaviours.

Furthermore, the study found a correlation between the leisure behaviour of older people and spatial elements such as the proportion of overhead coverage, the proportion of ground pavement, and functional diversity. This suggests that outdoor public spaces with canopies and large-scale hard pavements are more attractive to older people for leisure activities such as chess and card games. However, spaces with too many functional facilities can hinder such behaviours. This finding has not been reported in previous studies [30,41].

In addition, the study found that the number of fitness facilities and functional diversity are significantly related to the communication behaviour of older people. Diversity was found to be positively correlated with communication behaviour but negatively correlated with the number of fitness facilities. These findings are consistent with previous research [29]. The mixture of functions is conducive to promoting contact and communication among older people. However, spaces with a large amount of fitness equipment are not conducive to creating a stable communication atmosphere.

This study also found that the quality of exercise behaviour among older people was positively correlated with the number of fitness facilities, spatial connectivity, and plant diversity and negatively correlated with the green looking rate. These findings are consistent with previous research [29,31,78].

4.2. Interaction between Behavioural Motivation and Spatial Elements Configuration

The study found that there was an interaction and mutual restriction relationship between the behavioural motivation of older people and the configuration of spatial elements. In other words, there are common explanatory factors for behaviours under different goal orientations. For example, although the number of fitness facilities is negatively correlated with rest and communication behaviours, it is an important factor supporting the occurrence of exercise behaviours. Similarly, functional diversity is positively correlated with leisure behaviour and can also explain communication behaviour. This finding suggests that environmental conditions that support the occurrence of rest behaviours may not be able to accommodate the occurrence of exercise behaviours, even though both types of behaviours fall under the category of health promotion. Therefore, clarifying the relationship between spatial elements and various behavioural motivations is crucial for designing outdoor spaces that promote healthy behaviour among older people. This is the focus of this study and has not been discussed in previous research [42].

According to the interaction between behaviour motivation and the configuration of spatial elements, a health promotion-oriented model was created of the configuration of outdoor public space elements (Figure 10). As shown in Figure 10, the colour of each factor represents its positive or negative correlation to the type of behaviour, and the size of the circle’s radius represents the strength of the correlation. Correlation matching was performed on explanatory factors that appear multiple times in the matrix. First, if the two combinations have the same explanatory factors and the colours of the explanatory factors are consistent, then the environment has multiple possibilities for satisfying the two types of behaviours, and the combination modes of the two types of behaviour elements can be nested. Second, if the two behaviour combinations do not have the same explanatory factors, then the environment has the possibility of satisfying the two types of behaviours, and the two types of behaviour element combination modes are mutually compatible. Third, if the two combinations have the same explanatory factors but the colours of the explanatory factors are different, then the environment does not have the possibility of simultaneously satisfying the two types of behaviours, and the combination mode of the two types of behavioural elements is incompatible. Among these configurations, those that are mutually compatible and nestable can be combined and packaged together, while incompatible configurations must be packaged separately. Based on these principles, this model can not only reflect relatively clearly the configuration of spatial elements matching various exercise behaviours of older people but also reveal the nesting and compatibility relationships between the spatial elements.

4.3. Optimal Packing Combination Packages for Spatial Element Configuration

Due to the large differences in the requirements of the morphological characteristics of space for different health behaviour motivations of older people, there are various and complex combination patterns in the configuration of spatial elements. Moreover, due to the differences in site conditions, the arrangements and layout patterns of the existing residential areas in Shenzhen are not the same, which leads to different constraints on the design of outdoor public spaces in residential areas. This study proposes a list of residential outdoor public space configurations to meet diverse space needs. Based on the configuration model and the composite degree of health behaviour types, three-level packaging combinations (A, B, and C) were set up, and a packaging list was created (

Table 10. Optimal combination of residential outdoor public space element configurations.

Package ID	Priority	Supportive Behaviours	Positive List	Negative List
1	(A)	Oriented to promote rest and communication behaviours in older people	Ample space, diverse facilities, and visual focus are formed by a reasonable colour configuration.	Excess fitness equipment and facilities.
2	(B)	Oriented to promote the rest and leisure behaviours of older people	Ample space, spaces covered by a pavilion, sufficient hard pavement on the ground, and a visual focus formed by a reasonable colour arrangement.	Excessive fitness equipment and an overly complex combination of equipment and facilities.
3	(B)	Oriented to promote leisure and exercise behaviours of older people	Location of the site close to the centre of the residential area, a sufficient amount of fitness equipment, sufficient hard pavement, spaces covered by a pavilion, and a well-defined plant arrangement of high and low levels.	Excessive complex combination of fitness equipment and facilities and lush green plants.
4	(C)	Oriented to promote the rest behaviour in older people	Generous space and visual focus are formed by a reasonable colour configuration.	Excess fitness equipment and facilities.
5	(C)	Oriented to promote the leisure behaviour of older people	Spaces covered by a pavilion and sufficient hard pavement.	Excessive complex combination of props and facilities.
6	(C)	Oriented to promote the communication behaviour of older people	Diverse combination of facilities.	Excess fitness equipment and facilities.
7	(C)	Oriented to promote the exercise behaviour of older people	Location of the site close to the centre of the residential area, a sufficient amount of fitness equipment, and a well-defined plant arrangement of high and low levels.	Excessive lush green plants.

). The packaging rules are as follows:

(I)

Combinations of types A and B: the environment can satisfy and support healthy behaviours of two or more different motivations at the same time;

(II)

Type C combination: the environment can meet and support the healthy behaviour of at least one motivation.

When designing the design configuration according to the packing list, it is necessary to combine the specific site conditions of the outdoor public space. The following suggestions can be given: First, if the conditions of the space have inherent advantages—that is, they are not restricted by the configuration of various elements—priority can be given to selecting a level A or B spatial element packaging strategy that supports compound behaviours. This involves taking into account the health behaviour needs of older people for multiple purposes through the configuration of a single space to avoid the waste of space resources. If the existing site is relatively limited and some elements cannot be considered or configured, level C can be considered. In short, according to the characteristics of the spatial form of the residential area, the optimal packing list level is selected, a single or multi-purpose behavioural space is designed, the limitations of the spatial form are exploited, and the use efficiency of the outdoor public space is maximised.

4.4. Strengths and Limitations

Compared with other studies of the same type, this study has a certain degree of innovation in its research perspective, method, and conclusions. First, this study took the health promotion of older people as the anchor point to cut into the field of empirical research on the improvement of outdoor space quality in residential areas. It conducted a systematic discussion on the health behaviour of older people and the composition of micro-environmental elements to promote the construction and optimisation of a health promotion-oriented residential environment. Then, the extraction and quantification of behavioural data using the Mangold INTERACT comprehensive behavioural analysis platform replaced the traditional method of behavioural observation and recording, which provided ideas for subsequent variable subdivision in other similar studies. In terms of research conclusions, through rediscussing the subdivision of variables, this study attempts to construct an element configuration model and an optimal packing list for outdoor public space to meet diverse space needs. These outcomes make the design strategy more flexible and efficient, which was not available in previous studies.

Nevertheless, this study also has certain limitations. First, due to the workload, it was difficult to conduct research on the correlation between elements and behaviours over time. The data collection period was March and April (springtime in Shenzhen), and the sample data did not cover the entire year. In different seasons, older people’s use of space and their behavioural preferences may differ. It is hoped that future research will further discuss the differences in the correlation between environment and behaviour in the time dimension. Second, the Mangold INTERACT requires high-quality video for the behavioural data extraction. However, some videos captured by hidden cameras have low resolution, which may result in behavioural re-encoding errors. Finally, due to the limitation of the scene sample size, the sample types collected in this study could not cover all outdoor public spaces in residential areas in Shenzhen. The results are therefore only applicable to the boundary range of various types of indicators in the 40 samples, as shown in

Table A3. Indicator ranges for various types of environmental elements.

Environment Element	Indicator Range	Environment Element	Indicator Range	Environment Element	Indicator Range
1. Area	20–800	6. Green-looking Rate	18–75%	9. Recreational Facilities	0–16
2. Percentage of Ground Pavement	10–100%	7. Plant Diversity	0.51–2.76	10. Fitness Facilities	0–14
3. Percentage of Pavilion Coverage	0–100%	8. Colour Uniformity	0.25–0.89	11. Functional Diversity	0.32–1.44
4. Spatial Enclosure	0–91%
5. Spatial Connectivity	1–5

.

5. Conclusions

This study collected data on environmental composition indicators and behavioural activity among older people in outdoor public spaces in 40 residential areas in Shenzhen. An ordered logistic model was used to analyse the correlation between these two types of data. The results indicated that factors such as the scale of outdoor spaces, the size of hard ground areas, the quality of grey space, the green-looking rate, the accessibility of the site, the number of fitness facilities, and the diversity of site functions all have an impact on the outdoor health behaviour of older people. This suggests that the design of outdoor public spaces with a health-promotion orientation should take into account multiple factors, such as spatial composition, landscape composition, and facility composition.

Furthermore, unlike other similar studies, this study further subdivided behavioural variables using the Mangold INTERACT behavioural analysis platform. It was found that in complex combination modes, there are differences in the degree of support that environmental elements offer to multidimensional behaviours. Older people have different demands on space when resting, conversing, exercising, or engaging in leisure activities, which shows that adjusting the space elements according to local conditions can effectively promote the occurrence of healthy behaviours among older people. Based on this analysis, an optimal configuration model for health promotion in residential outdoor spaces was constructed. The model proposes a multi-adaptive level configuration list to address outdoor public spaces at different levels within residential areas. The findings provide a scientific and effective strategy for optimising the quality of outdoor environments. In addition to Shenzhen, countries such as South Korea, Japan, Singapore, and Brazil also have large-scale congregate housing. The findings can provide a more scientific and effective strategy for optimising the quality of outdoor public spaces in such residential areas.

Prior to the incorporation of health behaviour theory, this research on outdoor environments focused on the overall planning and design of residential areas. Few studies have examined the correlation between activity characteristics and the adaptive needs of older people in outdoor spatial environments. Adhering to the principle of promoting healthy outdoor activities, this research analysed and refined the characteristics of health promotion in outdoor activity spaces. It expands the practical application of health promotion theory by enhancing the development of outdoor environments for ageing residential areas. More importantly, this study demonstrates a new research methodology for environmental analysis based on objective behavioural surveys. Within the framework of traditional empirical research, this study attempts to introduce the video behaviour analysis software Mangold INTERACT from the fields of psychology and cognitive neuroscience. Through specific research cases, it is shown that the iteration of behavioural survey tools can expand the research depth of such surveys. It is believed that this can provide a reference for other quantitative research on environment and behaviour. Nevertheless, this study has limitations, such as a relatively restrictive time period for sampling behaviours, low-quality sampled videos, and insufficient sample coverage. Future research could benefit from introducing time variables and using a larger sample size to further explore the correlation between environment and health behaviour.