Abstract
Background
Shift work is essential for nurses and is the backbone of the healthcare workforce. Addressing the challenges associated with time-consuming scheduling is crucial for ensuring nurses’ work quality, optimal staffing levels, and increased job satisfaction. We compared the work quality from both organizational and individual perspectives after the implementation of the Inha University Hospital Nursing Artificial Intelligence (AI) Scheduling System (IH-NASS), and analyzed the factors influencing nurses’ job satisfaction, focusing on their perceptions of IH-NASS and work quality.
Methods
A total of 253 shift nurses from 14 wards where the IH-NASS was implemented at a tertiary university hospital in Korea were selected. Data from the traditional manual (December 2022, retrospective study) and the IH-NASS-generated schedules (December 2023, prospective study) were compared. Nurses’ general characteristics, IH-NASS perceptions (convenience, satisfaction, and fairness), and job satisfaction were surveyed and analyzed.
Results
Compared to traditional manual schedules, IH-NASS-generated schedules significantly reduced the number of nurses with < 1 year of experience in day shifts. From an individual perspective, the number of night-off-evening (NOE) shifts was significantly lower. Additionally, IH-NASS-generated schedules had more consecutive off days (≥ 2), off days (≥ 2) following two or more consecutive night shifts, Saturday-Sunday off days, and Sunday off days, whereas weekday shifts with unsocial hours were fewer. Factors influencing job satisfaction among shift nurses included satisfaction with the IH-NASS, perceived convenience of the IH-NASS, and the number of NOE shifts under unhealthy work scheduling, which together accounted for approximately 27% of the variance in job satisfaction.
Conclusions
This study provides empirical evidence supporting the use of AI systems in nurse scheduling. Specifically, AI-based scheduling can optimize workforce allocation while maintaining work quality, enhancing nurses’ positive perceptions, and improving job satisfaction.
Trial registration
Not applicable. This was not a clinical trial.
Keywords: Shift work schedule, Nurse, Perception, Job satisfaction, Artificial intelligence
Introduction
South Korea had 503,831 licensed nurses in 2023, with 243,325 of them working in hospitals [1]. Notably, less than 50% of licensed nurses are employed in medical institutions. The total number of clinical nurses per 1,000 people was 9.1, lower than the Organization for Economic Cooperation and Development (OECD) average of 9.7 [2]. Compared to major developed countries, the Korean workforce remains insufficient. Therefore, ensuring a stable nursing workforce is essential.
Shift work is essential in nursing because of the need for continuous patient care [3, 4]. Nurse shift scheduling can affect staff mental and physical health [5], job satisfaction [6], and staff retention [7]. Finally, the scheduling of physicians and nurses in response to a constantly changing number of patients directly affects hospital efficiency and patient satisfaction [8]. Notably, nurse shift scheduling in hospitals is manually conducted by unit nurse managers. This process is time consuming and prone to overlooking work characteristics [5, 6, 9, 10]. Traditional manual scheduling has limitations in ensuring balanced and fair schedules while accommodating individual needs [8]. Consequently, scheduling and various nurse-scheduling problems have become research hotspots in recent years.
Previous studies have proposed scheduling strategies that maximize job satisfaction by incorporating impartiality in workload distribution and individual shift preferences [11]. Studies show that compared to traditional manual scheduling, participatory work-time scheduling software has been associated with modest changes in work-time characteristics and with greater control over shift schedules, reduced excessive sleepiness [7], and improved sleep and work performance [12]. Participatory working time scheduling software has been used to help healthcare professionals organize their schedules based on rules and principles, such as ensuring a minimum number of weekends and night shifts [7, 12, 13], and preferences for the number of hours per week and start times [8]. Rerkjirattikal et al. [6] developed a goal-programming approach to nurse scheduling that integrates factors related to job satisfaction, demonstrating improved perceptions of fairness, preferences, and satisfaction among nurses. Electronic and self-scheduling systems have been shown to have positive effects at individual and organizational levels [14, 15].
In nursing workforce research [16], the organizational perspective includes team-level characteristics and structural factors such as adequate staffing and appropriate scheduling, and the adequateness of working conditions, which contribute to work quality. Assigning nurses to the right departments and shifts based on their human factors of skills, preferences, and suitability can strengthen teamwork and increase healthcare system efficiency [17]. In particular, at the organizational level, unlike shift scheduling in other professions, nurse scheduling must account for differences in the scope of responsibilities and task acceptance based on clinical experience and proficiency. Therefore, allocating teams with a balanced mix of novice, competent, and proficient nurses is essential to ensure patient safety and effective responses to emergencies [3, 16]. Proficient nurses are capable of managing complex clinical situations and mentoring junior staff, competent nurses can perform routine clinical tasks independently, and novice nurses generally require supervision when performing duties [18]. Shift teams with well-balanced proficiency levels have been shown to distribute tasks more efficiently and reduce the burden of training and supervision [19, 20]. In this study, the proficiency level was set as a key indicator of work quality within an organization. Based on Benner’s five-stage theory [18] and reflecting actual hospital staffing standards, nurses were categorized into following three levels: expert, competent, and novice. Using these classifications, the artificial intelligence (AI) system automatically generated shift schedules to ensure work quality and appropriate workforce allocation.
In this study, the Inha University Hospital Nursing Artificial Intelligence Scheduling System (IH-NASS), an optimization-based approach designed to overcome the limitations of traditional manual scheduling and enhance job satisfaction, was implemented to enhance individual flexibility and work quality during nurse shifts. Integrating AI into the healthcare system can transform and optimize processes [21–24]. In addition, in the era of intelligent data, full utilization can realize reform and innovation in hospital human resource management and enhance the advantages of staff [23, 24]. This AI system incorporates working conditions and types, number of nurses per shift, departmental constraints, and individual nurse preferences to ensure that the human factors identified in previous studies, such as nurses’ needs, preferences, and workload fairness, are met.
Overall, in this study, we compared shift nurses’ work quality from both organizational and individual perspectives after implementation of the IH-NASS, and examined the factors influencing nurses’ job satisfaction, including their perceptions of the IH-NASS and work quality.
Methods
Study design
In this descriptive study, using a pre-post comparative design with an analytical approach, we compared shift nurses’ work quality before and after IH-NASS implementation, and analyzed the factors influencing nurses’ job satisfaction.
Research setting and sample
This study was designed to analyze the organizational and individual perspectives of the IH-NASS. An organizational perspective analysis was conducted on 14 wards implementing the IH-NASS at Inha University Hospital in a metropolitan city with > 30 million people near Seoul, Korea. It is a tertiary university hospital with approximately 900 beds, 35 medical departments, and > 1,200 nurses.
The individual-perspective sample included shift nurses working in the 14 wards of the IH-NASS. The inclusion criteria for shift nurses were as follows: (1) working in a three-shift system, (2) participating in the direct care of inpatients, and (3) who had been working in the same ward from December 2022, when traditional manual scheduling was applied, to December 2023, when the IH-NASS was implemented. The exclusion criteria were as follows: (1) nurses who were not subject to the three-shift system, including night-duty nurses, fixed-shift nurses, or those with specific work exclusions due to personal reasons such as pregnancy; and (2) nurses who did not provide direct care for inpatients, such as nurse managers and dedicated clinical practice nurses. The sample size required for this study was determined using the G*Power 3 program [25]. With a significance level of 0.05, medium effect size of 0.15, power of 0.90, and 30 predictors (general characteristics with 11 items, work quality with 19 items, and perception of IH-NASS with 3 items), the computed target sample size was 235. Considering a dropout rate of approximately 10%, 261 participants were required for the study. Through the Nursing Department, a list of nurses working as of December 1, 2022, and December 1, 2023, was obtained. In total, 259 nurses working in the same general ward were matched during both periods. The researcher visited 14 general wards during handover times to explain the study’s purpose and the inclusion and exclusion criteria to nurse managers and shift nurses, and recruited voluntary participants. Five were excluded due to nonresponse, leaving 253 shift nurses in the final analysis, eventually meeting the sample size.
Measurements
General characteristics of participants
We recorded the shift nurses’ sex, age, education level, total clinical experience, current departmental experience, marital status, monthly income, perceived health, sleep duration, insomnia, and excessive sleepiness over the last 3 months.
Work quality
Work quality is a broad concept encompassing multiple dimensions at both the individual and organizational levels. In this study, work quality was analyzed using specific indicators from these two perspectives. From an organizational perspective, nurse proficiency, which typically improve with clinical experience, have been identified as key factors influencing the quality of patient care within nursing teams [20]. Accordingly, this study simplified Benner’s five-stage theory [18] into three levels—novice (< 1 year), competent (1–3 years), and proficient (> 3 years)—based on clinical experience, thereby reflecting the practical criteria used for staffing and shifting team allocation at Inha University Hospital. Using these categories, the traditional manual schedule (December 2022) was compared to IH‑NASS–generated schedules (December 2023).
From an individual perspective, work quality was assessed using objective schedule quality indicators that reflected the structural characteristics of work schedules, based on prior studies [7, 11, 13, 26]. First, the number of workdays for each shift type (day [06:30–15:30], evening [14:30–23:30], night [22:30–07:30], and off) and total working hours were measured. Second, unhealthy work scheduling was measured based on the definition provided by a previous study [19] as follows: “schedules that do not guarantee minimum rest time or disrupt circadian rhythms, making it extremely difficult for nurses to work.” Accordingly, the number of night-off-day (NOD), night-off-evening (NOE), evening day (ED), and evening off-day (EOD) shifts were assessed. Third, the number of instances of two or more consecutive off days was measured. Fourth, the number of consecutive night shifts was assessed. Fifth, the number of instances of two or more consecutive off days following two or more consecutive night shifts was measured. Sixth, the number of off days on Saturdays and Sundays was measured. Seventh, unsocial working hours were assessed based on the 2022 National Health Service [27] criteria in the United Kingdom, defined as weekdays from 8:00 PM to 6:00 AM, Saturdays from 12:00 AM to 11:59 PM, and Sundays and public holidays from 12:00 AM to 11:59 PM.
Perception of the IH-NASS
The perception of the IH-NASS was assessed using questionnaire items developed based on prior studies [6, 14, 26, 28], indicating that subjective schedule quality plays a significant role in evaluating scheduling systems. The following three aspects were examined using a four-point Likert scale: (1) convenience of shift requests through the IH-NASS, (2) satisfaction with the IH-NASS-generated schedules, and (3) fairness of that month’s schedule regarding workload distribution, preferred shifts and holidays, number of night shifts, and requested days off. Higher scores indicated a more positive perception of the IH-NASS. The Cronbach’s α for this study was 0.76.
Job satisfaction
We used a standardized questionnaire on job satisfaction from The Korean Longitudinal Survey of Women and Families [29]. Ten questions were asked using a five-point Likert scale, with higher scores indicating higher job satisfaction. The Cronbach’s ⍺ of the scale in a previous study [30] was 0.91 compared to 0.89 in this study.
Application of the IH-NASS
The IH-NASS is an AI-driven nurse scheduling system developed from April to October 2022 by a team consisting of seven nurses with > 17 years of clinical experience and two information technology (IT) specialists, in collaboration with the supplier, Kaiem Co., Ltd. During development, hospital-specific shift codes, integration with the hospital’s IT system, and optimization of work conditions were incorporated into the AI algorithm.
This AI system automatically generates optimal schedules by reflecting various conditions, including a guaranteed minimum 14-h rest period between shifts, providing two off-days after five consecutive workdays, assigning a minimum number of team members per shift, balancing nurse proficiency levels by preventing less experienced nurses from being assigned to the same team, and implementing scheduling rules designed to improve nurse satisfaction. It also reduces double data entry by directly linking the finalized schedules to the hospital’s workforce management system, thereby enhancing user convenience.
Training sessions for nurse managers will be conducted four times from late July to late September 2022. After testing the AI program, feedback was collected and used to improve the system. The process worked as follows: nurses submit their preferred shifts; nurse managers set the necessary scheduling parameters, including staff information and scheduling constraints; AI automatically generates optimized schedules; nurse managers finalize and distribute the schedules to nurses; and the finalized schedules are integrated with the hospital’s IT system.
Although the system was implemented in October 2023, manual scheduling was still used initially because of unfamiliarity with the program. Through one-on-one coaching and manager meetings, the nurse managers’ proficiency in the AI system improved, leading to stable adoption of the IH-NASS across departments.
Data collection
Data from the traditional manual (December 2022, retrospective study) and IH-NASS-generated schedules (December 2023, prospective study) were used to compare the aspects before and after implementation of the IH-NASS. From December 4 to December 20, 2023, copies of the traditional manual scheduling from December 2022 were collected by the nurse managers of the 14 wards. The IH-NASS-generated schedules for December 2023 were retrieved in bulk on December 4 via the institutional Human Resource Management System. The collected schedules were manually counted and coded according to predefined criteria by four research assistants, and the data were subsequently reviewed by the researcher to ensure reliability. A questionnaire on the participants’ general characteristics, perceptions of the IH-NASS (convenience, satisfaction, and fairness), and job satisfaction was administered between December 5 and December 20, 2023.
Data analysis
Statistical analyses were conducted using SPSS/WIN 28.0. First, the participants’ general characteristics, perceptions of the IH-NASS (convenience, satisfaction, and fairness), and job satisfaction were analyzed using frequencies, percentages, means, and standard deviations. Second, differences in job satisfaction according to participants’ general characteristics were analyzed using an independent t-test, one-way analysis of variance, and Scheffé post-hoc tests. Subsequently, variables showing statistically significant differences were included in a multiple regression analysis to identify the factors influencing job satisfaction. Third, work quality (organizational and individual perspectives) was compared between the traditional manual and IH-NASS-generated schedules using a paired t-test. Fourth, factors influencing job satisfaction were analyzed using multiple regression analysis.
Ethical considerations
The research protocol was approved by the Institutional Review Board (IRB No. 2023-11-037) of the principal investigator’s institution prior to data collection. Participants were included in the study if they had fully read the informed consent form and voluntarily agreed to participate.
Results
Differences in job satisfaction according to participants’ general characteristics
Of 253 shift nurses, 98.0% were female. The mean age was 27.82 ± 4.31 years (min: 23, max: 55). Most participants had a bachelor’s degree (91.7%). The total clinical experience as a nurse was 59.16 ± 52.82 months (min: 12, max: 374). The current departmental experience was 44.91 ± 29.67 months (min: 12, max: 209). In terms of marital status, 85.8% of the participants were single. The most common monthly income range was between 3 million and 4.5 million KRW, accounting for 57.9% of participants. Job satisfaction among nurses significantly differed by educational level (F = 4.44, p = 0.013), perceived health status (F = 3.08, p = 0.017), sleep duration (t = − 2.43, p = 0.008), insomnia in the last 3 months (F = 6.68, p < 0.001), and excessive sleepiness in the last 3 months (F = 12.13, p < 0.001) (Table 1).
Table 1.
Differences in job satisfaction according to shift nurses’ general characteristics (n = 253)
| Characteristics | Categories | n (%) or Mean ± SD | Job satisfaction | |
|---|---|---|---|---|
| Mean ± SD | t of F(p) | |||
| Sex | Man | 5 (1.9) | 32.60 ± 5.59 | 0.96 (0.337) |
| Woman | 248 (98.0) | 29.89 ± 6.25 | ||
| Age (in years) | (min: 23, max: 55) | 27.82 ± 4.31 | ||
| 21 to 30 | 214 (84.6) | 29.81 ± 6.09 | 0.71 (0.495) | |
| 31 to 40 | 32 (12.6) | 30.25 ± 7.02 | ||
| ≧ 41 | 7 (2.8) | 32.57 ± 7.25 | ||
| Education level | Diplomaa | 13 (5.1) | 28.92 ± 3.29 | 4.44 (0.013) |
| Bachelorb | 232 (91.7) | 30.21 ± 6.29 | b > c | |
| ≧Masterc | 8 (3.2) | 23.75 ± 4.83 | ||
| Total clinical experience (month) | (min: 12, max: 374) | 59.16 ± 52.82 | ||
| Current departmental experience (months) | (min: 12, max: 209) | 44.91 ± 29.67 | ||
| Marital status | Married | 33 (13.0) | 28.64 ± 6.80 | 1.41 (0.247) |
| Single | 217 (85.8) | 30.19 ± 6.11 | ||
| Divorce | 3 (1.2) | 26.33 ± 9.02 | ||
| Monthly income (KRW) (n = 252) | 1.5–3.0 million | 22 (8.7) | 28.64 ± 6.00 | 1.49 (0.204) |
| 3.0–4.5 million | 146 (57.9) | 29.74 ± 6.48 | ||
| 4.5–6.0 million | 19 (7.5) | 31.42 ± 5.34 | ||
| 6.0–7.5 million | 22 (8.7) | 28.59 ± 5.65 | ||
| 7.5 million or more | 43 (17.1) | 31.51 ± 5.99 | ||
| Perceived health status (n = 252) | Very bada | 3 (1.2) | 27.33 ± 5.51 | 3.08 (0.017) |
| Badb | 39 (15.5) | 27.92 ± 7.59 | d > b | |
| Normalc | 143 (56.7) | 29.65 ± 5.87 | ||
| Goodd | 61 (24.2) | 32.07 ± 5.94 | ||
| Very goode | 6 (2.4) | 29.67 ± 3.88 | ||
| Sleep duration (hours) | (min: 2, max: 10) | 6.54 ± 1.33 | ||
| ≤ 6.9 | 141 (55.7) | 29.10 ± 6.85 | -2.43 (0.008) | |
| ≥ 7 | 112 (44.3) | 31.00 ± 5.22 | ||
| Insomnia (last 3 months) | Not at alla | 19 (7.5) | 32.00 ± 4.36 | 6.68 (< 0.001) |
| Sometimesb | 120 (47.4) | 30.72 ± 6.19 | a > b, c, d | |
| Frequentc | 76 (30.0) | 30.16 ± 6.25 | ||
| Very oftend | 38 (15.0) | 26.05 ± 5.81 | ||
| Excessive sleepiness (last 3 months) | Not at all | 18 (7.1) | 32.39 ± 6.42 | 12.13 (< 0.001) |
| Sometimes | 94 (37.2) | 31.77 ± 5.37 | a > b, c, d | |
| Frequent | 93 (36.8) | 29.77 ± 6.28 | ||
| Very often | 48 (19.0) | 25.79 ± 5.74 | ||
KRW: Korean Won
Organizational perspective: comparison of shift-based clinical experience allocation between traditional manual scheduling and IH-NASS
The IH-NASS-generated schedules had a significantly lower number of novice nurses (< 1 year of experience) on day shifts than the traditional manual schedules (t = 2.70, p = 0.018).
Individual perspectives: perceptions of the IH-NASS and job satisfaction
Shift nurses’ perceptions of the IH-NASS were 2.54 ± 0.73 for convenience, 2.12 ± 0.74 for satisfaction, and 2.44 ± 0.73 for fairness on a four-point scale. Job satisfaction had a mean of 29.94 ± 6.24 out of 45 (minimum: 10, maximum: 45) (Table 2).
Table 2.
Comparison of shift-based clinical experience allocation between traditional manual scheduling and IH-NASS (n = 14)
| Categories | Traditional manual scheduling (2022.12) | IH-NASS (2023.12) |
Difference (2022–2023) |
t (p) |
|---|---|---|---|---|
| Mean ± SD | ||||
| Day shift | ||||
| < 1 year | 1.43 ± 0.37 | 1.00 ± 0.51 | 0.43 ± 0.59 | 2.70 (0.018) |
| < 3 years | 1.73 ± 0.88 | 1.80 ± 0.51 | -0.06 ± 0.77 | − 0.29 (0.776) |
| ≥ 3 years | 3.44 ± 0.49 | 3.65 ± 1.13 | -0.21 ± 1.08 | − 0.73 (0.481) |
| Evening shift | ||||
| < 1 year | 1.32 ± 0.38 | 1.04 ± 0.55 | 0.28 ± 0.56 | 1.87 (0.084) |
| < 3 years | 1.91 ± 0.78 | 1.96 ± 0.50 | -0.05 ± 0.66 | − 0.30 (0.769) |
| ≥ 3 years | 3.35 ± 0.37 | 3.64 ± 1.21 | -0.29 ± 1.19 | − 0.901 (0.384) |
| Night shift | ||||
| < 1 year | 1.12 ± 0.33 | 0.93 ± 0.47 | 0.19 ± 0.54 | 1.33 (0.205) |
| < 3 years | 1.75 ± 0.75 | 1.97 ± 0.56 | -0.22 ± 0.53 | -1.53 (0.149) |
| ≥ 3 years | 3.07 ± 0.50 | 3.20 ± 1.03 | -0.13 ± 0.89 | − 0.55 (0.590) |
IH-NASS: Inha University Hospital Nursing AI Scheduling System
Individual perspective: comparison of work quality between traditional manual scheduling and the IH-NASS
Compared to traditional manual schedules, IH-NASS-generated schedules had a significantly lower number of NOE shifts, a type of unhealthy work scheduling (t = 2.64, p = 0.009). The number of consecutive off days (≥ 2) (t = − 3.78, p < 0.001), off days (≥ 2) following consecutive night shifts (≥ 2) (t = − 2.13, p = 0.034), off days on both Saturday and Sunday (t = − 2.24, p = 0.026), and off days on Sunday (t = − 3.50, p < 0.001) was significantly greater. Weekday unsocial hours were significantly lower (t = 2.74, p = 0.007), while unsocial hours on Saturdays and Sundays were significantly greater (t = − 2.85, p = 0.005) (Table 3).
Table 3.
Comparison of work quality between traditional manual scheduling and IH-NASS (n = 253)
| Categories | Handwritten timesheets (2022.12) |
IH-NASS (2023.12) |
Difference (2022–2023) |
t (p) |
|---|---|---|---|---|
| Mean ± SD | ||||
| Number of day shifts | 6.50 ± 2.25 | 6.13 ± 2.26 | 0.37 ± 3.21 | 1.82 (0.070) |
| Number of evening shifts | 6.81 ± 2.23 | 6.89 ± 2.18 | -0.08 ± 3.11 | − 0.41 (0.686) |
| Number of night shifts | 5.36 ± 1.14 | 5.46 ± 1.07 | -0.09 ± 1.12 | -1.35 (0.177) |
| Number of off days | 11.23 ± 1.34 | 11.40 ± 0.96 | -0.17 ± 1.53 | -1.7 (0.078) |
| Total working hours | 164.39 ± 11.79 | 165.73 ± 7.89 | -1.34 ± 13.90 | -1.54 (0.125) |
| Unhealthy work scheduling | ||||
| Number of NOD | 0 | 0 | ||
| Number of NOE | 0.49 ± 0.66 | 0.34 ± 0.61 | 0.15 ± 0.90 | 2.64 (0.009) |
| Number of ED | 0 | 0 | ||
| Number of EOD | 0.78 ± 0.79 | 0.66 ± 0.76 | 0.11 ± 1.00 | 1.76 (0.080) |
| Number of consecutive off days (≥ 2) | 3.06 ± 0.95 | 3.35 ± 0.93 | -0.29 ± 1.23 | -3.78 (< 0.001) |
| Number of consecutive night shifts | 2.10 ± 0.63 | 2.15 ± 0.59 | -0.06 ± 0.76 | -1.16 (0.247) |
| Number of off days (≥ 2) after consecutive night shifts (≥ 2) | 1.53 ± 0.77 | 1.66 ± 0.75 | -0.14 ± 1.03 | -2.13 (0.034) |
| Number of off days on Saturdays and Sundays | 3.74 ± 1.70 | 4.04 ± 1.56 | -0.29 ± 2.04 | -2.24 (0.026) |
| Number of off days on Saturdays | 2.05 ± 1.13 | 2.04 ± 1.04 | 0.01 ± 1.36 | 0.09 (0.927) |
| Number of off days on Sundays | 1.70 ± 1.00 | 1.99 ± 1.03 | -0.30 ± 1.35 | -3.50 (< 0.001) |
| Total unsocial hours on weekdays | 39.49 ± 10.05 | 37.11 ± 9.91 | 2.39 ± 13.86 | 2.74 (0.007) |
| Total unsocial hours on Saturdays and Sundays | 46.87 ± 14.85 | 49.92 ± 12.07 | -3.06 ± 17.06 | -2.85 (0.005) |
| Total unsocial hours | 86.36 ± 17.38 | 87.03 ± 13.70 | -0.67 ± 20.76 | -0.52 (0.607) |
| Proportion of unsocial hours | 0.53 ± 0.10 | 0.53 ± 0.08 | -0.0006 ± 0.12 | -0.07 (0.944) |
IH-NASS: Inha University Hospital Nursing AI Scheduling System, NOD: night-off-day; NOE: night-off-evening, ED: evening-day, EOD: evening-off-day
Individual perspective: factors influencing the job satisfaction of shift nurses after IH-NASS implementation
A multiple regression analysis was conducted to examine the factors influencing job satisfaction among shift nurses after the IH-NASS implementation in December 2023, including shift nurses’ work quality and general characteristics. The Durbin–Watson value was 2.01, indicating that the residuals were not autocorrelated. The tolerance ranged 0.52–0.98, which is > 0.1, and the variance inflation factor (VIF) ranged 1.02– 2.02, which is not > 10. Thus, multicollinearity among the independent variables was not a challenge.
The regression model, which included general characteristics that showed statistically significant differences in job satisfaction—education level, perceived health status, sleep duration, insomnia in the last 3 months, excessive sleepiness in the last 3 months—along with perception of IH-NASS (convenience, satisfaction, and fairness), and work quality as independent variables, was statistically significant (F = 12.44, p < 0.001). After IH-NASS implementation, the factors influencing shift nurses’ job satisfaction included excessive sleepiness, education level (university), sleep duration (≥ 7 h), satisfaction with IH-NASS (β = 0.24, p < 0.001), convenience of IH-NASS (β = 0.16, p = 0.016), and number of NOE shifts (β = 0.12, p = 0.033). This model explains approximately 27% of the variance (Table 4).
Table 4.
Factors influencing shift nurses’ job satisfaction after IH-NASS implementation (n = 253)
| Variables | Job satisfaction | |||||
|---|---|---|---|---|---|---|
| B | SE | β | t | p | 95% CI | |
| (Constant) | 15.83 | 1.73 | 9.15 | < 0.001 | 12.42–19.23 | |
| Excessive sleepiness (last 3 months) * (1 = sometimes) | 4.12 | 0.99 | 0.32 | 4.17 | < 0.001 | 2.18–6.07 |
| Satisfaction with IH-NASS | 2.05 | 0.56 | 0.24 | 3.66 | < 0.001 | 0.94–3.15 |
| Excessive sleepiness (last 3 months) * (1 = not at all) | 5.06 | 1.51 | 0.21 | 3.35 | 0.001 | 2.09–8.04 |
| Excessive sleepiness (last 3 months) * (1 = frequent) | 2.69 | 0.97 | 0.21 | 2.77 | 0.006 | 0.77–4.60 |
| Convenience for IH-NASS | 1.37 | 0.57 | 0.16 | 2.42 | 0.016 | 0.26–2.48 |
| Number of NOE | 1.20 | 0.56 | 0.12 | 2.15 | 0.033 | 0.10–2.31 |
| Education* (1 = university) | 2.63 | 1.26 | 0.11 | 2.09 | 0.038 | 0.15–5.10 |
| Sleep duration* (1 = 7 h or more) | 1.38 | 0.69 | 0.11 | 2.01 | 0.045 | 0.03–2.73 |
| R2 = 0.29, Adj. R2 = 0.27, F = 12.44, p < 0.001 | ||||||
IH-NASS: Inha University Hospital Nursing AI Scheduling System. *Dummy variables; NOE: Night-off-evening
Discussion
In this study, we compared shift nurses’ work quality from both organizational and individual perspectives after IH-NASS implementation, and identified the factors influencing nurses’ job satisfaction, including perceptions of IH-NASS and work quality. Overall, this study provides valuable empirical data that can be used for human resource management in the nursing field by utilizing an AI system for nurse work assignments and analyzing its effects.
From an organizational perspective, an analysis of 14 departments revealed that compared to traditional manual schedules, IH-NASS-generated departments had significantly fewer novice nurses (with < 1 year of experience) assigned to day shifts, during which major treatment plans were implemented. This is an important finding, indicating that the IH-NASS is an effective tool for efficiently staffing nurses according to proficiency level, even at the departmental level. Meanwhile, participatory working time scheduling software has a disadvantage in that each nurse only applies their own schedule, resulting in shift imbalances with colleagues [15]. A literature review of new nurses’ patient safety and practice knowledge [31] found that new nurses have different levels of knowledge and practice competence depending on their experience, which can pose a threat to patient safety. The responsibility for patient safety should not be limited to the nurse’s practice; it is the responsibility of all members of the healthcare system [31]. Therefore, strategies for establishing efficient and balanced shift organizations are essential to ensure patient safety [32]. We established a system through the IH-NASS that can help ensure quality nursing care in the department by assigning nurses with nursing competencies to each shift and ensuring a safe continuity of care.
From an individual perspective, shift nurses had a relatively positive perception of IH-NASS, with convenience being rated the highest, followed by fairness and satisfaction. The AI system increased convenience, which is expected to provide more opportunities to focus on patient care in clinical practice [24]. Positive perceptions of nurse schedules are essential factors in nurse retention [4]. Shift scheduling is largely performed manually, which results in significant time and resource inefficiencies [5, 6, 9, 10]. When multiple nurses request personal preferred work schedules, nurse managers are likely to be influenced by subjective factors, may not be able to meet al.l nurses’ needs fairly, and may experience staff conflict [23, 33]. In addition, qualitative research on sustainable working time management emphasizes the importance of fairly distributing unpopular shifts, such as evening, night, weekend shifts, and public holidays [3]). Similarly, a study on the implementation of an AI-generated self-rostering system for radiology staff in New Zealand expressed concerns about its implementation and uncertainty about its fairness. However, it could be fairer than manual scheduling and save time and money [21].
From an individual perspective, compared to traditional manual schedules, IH-NASS schedules have significantly fewer NOE shifts, which is a type of unhealthy work scheduling. Additionally, there were more consecutive off days (≥ 2), off days (≥ 2) following consecutive night shifts (≥ 2), off days on both Saturday and Sunday, and off days on Sunday. Meanwhile, the number of unsocial hours on weekdays decreased. Similar to our findings, another study showed that a goal programming approach to nurse scheduling for operating room nurses in Thailand, which simultaneously considered workload fairness and individual preferences, yielded a more balanced workload allocation compared to traditional manual scheduling when preferred shifts and off days were considered [6]. A South Korean study [4] found that 68.0% of nurses did not receive a guaranteed 48-h rest period after two or more consecutive night shifts. Unhealthy work scheduling is a work schedule that poses a threat to nurses’ health, including schedules that do not allow sufficient rest between consecutive shifts, contrary to circadian rhythms, which involve more than five consecutive days [16], and cause social jet lag [34]. Socially jet-lagged work schedules, such as NOD and NOE, result in a mismatch between nurses’ circadian rhythms and work schedules [34] and require physical adaptations that significantly burden nurses [16]. Meanwhile, the IH-NASS uses an AI system that can efficiently derive an optimized work schedule that minimizes unhealthy work scheduling and unsocial work hours and secures rest periods within a limited workforce without increasing the workload.
In contrast, the IH-NASS schedules had significantly more unsocial working hours on Saturdays and Sundays than the traditional manual schedules. As night and holiday shifts cause work-life imbalances [35], nurses should allocate unsocial hours to minimize their negative effects [4]. Unsocial working hours cannot be avoided because of the nature of nurses’ work, which involves working shifts of 24 consecutive hours [4, 36]. Nevertheless, this aspect should be improved and monitored by utilizing the working condition value-setting function in the IH-NASS. The factors that influenced nurses’ job satisfaction after IH-NASS implementation were satisfaction and convenience among the perceptions of the IH-NASS and the number of NOE shifts among unhealthy work scheduling, with an explanatory power of approximately 27%. Nurses’ job satisfaction is mainly determined by their working hours and is an important factor in their quality of work life [37]. Given that nurses’ job satisfaction is associated with organizational productivity variables, such as nurse retention [38], our findings provide potential benefits for healthcare institutions facing challenges in managing and retaining nursing staff by examining the impact of IH-NASS implementation on shift nurses’ job satisfaction. It establishes a foundation for the broader adoption of the system, and is therefore significant.
This study has some limitations. First, we analyzed data from a single tertiary university hospital that applied the IH-NASS. Therefore, these results should be interpreted with caution. Similar studies should be conducted in various countries, regions, healthcare institutions, and bed sizes. Second, although the measurement tools used in this study were developed based on prior literature, their validity and reliability have not yet been established, which may represent a limitation of this study. Therefore, future research should focus on developing instruments with verified reliability and validity to reexamine the findings of this study. Additionally, qualitative research is warranted to address the limitations of current instruments and gain a deeper understanding of the experiences of both nurse managers and shift nurses. Third, the IH-NASS function was established and applied considering individual nurses and departments; however, it did not reflect patient severity or bed utilization rates. As AI advances, the development of an optimized algorithmic system that utilizes patient and hospital data to solve this problem is an interesting topic for future research.
Conclusions
From an organizational perspective, this study found that after implementing the IH-NASS, compared to traditional manual schedules, the number of novice nurses assigned to day shifts decreased. From individual nurses’ perspectives, insufficient rest periods between consecutive shifts decreased, consecutive off-days increased, and unsocial hours on weekdays decreased. Shift nurses had relatively positive perceptions of the IH-NASS. Additionally, among the IH-NASS perceptions, the number of NOE shifts, satisfaction, and convenience were identified as significant factors that positively influenced nurses’ job satisfaction. Thus, applying AI systems in scheduling can enable shift allocation that maintains work quality at both the individual and organizational levels, while enhancing nurses’ positive perceptions and job satisfaction. Further large-scale and longitudinal studies are needed to accumulate evidence for the successful implementation of AI systems in nursing scheduling. Scholars should extract diverse insights into the positive impacts of AI-based nurse scheduling, in terms of nursing outcomes and those of improvements in patient care quality and outcomes, workforce planning in healthcare institutions, and healthcare expenditure.
Acknowledgements
The Inha University Hospital Nursing AI Scheduling System (IH-NASS) was introduced with support from the 2022 AI Voucher Support Program, supervised by the National IT Industry Promotion Agency (NIPA).
Abbreviations
- AI
Artificial intelligence
- ED
Evening-day
- EOD
Evening-off-day
- IH-NASS
Inha university hospital nursing artificial intelligence scheduling system
- IRB
Institutional review board
- IT
Information technology
- KRW
Korean won
- NIPA
National IT industry promotion agency
- NOD
Night-off-day
- NOE
Night-off-evening
- OECD
Organization for economic cooperation and development
- VIF
Variance inflation factor
Author contributions
HWK, JK, KK, EKB, HK, JHJ, and WC conceived and designed the study. HWK and JK collected and analyzed the data. JK was a major contributor to the manuscript writing. JK prepared Tables 1, 2 and 3, and 4. All authors have reviewed, read, and approved the final manuscript.
Funding
None.
Data availability
The data that support the findings of this study are available from the corresponding author upon reasonable request. However, restrictions apply to the availability of these data, which contain sensitive institutional information and identifiable personnel data. Therefore, they are not publicly available. Data access may be granted upon request and with permission from Inha University Hospital.
Declarations
Ethical approval and consent to participate
This study was approved by the Institutional Review Board of Inha University Hospital (No. 2023-11-037). All participants provided written informed consent to participate in the research in accordance with the Declaration of Helsinki. All methods were performed in accordance with the relevant guidelines and regulations in the Ethical Declarations.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Ministry of health and welfare. Health and welfare statistical yearbook. 2024; p. 09. https://www.mohw.go.kr/board.es?mid=a10411010100%26bid=0019%26tag=%26act=view%26list_no=14841.
- 2.Organization for economic coperation and development (OECD). OECD health statistics 2024. https://www.oecd.org/en/data/datasets/oecd-health-statistics.html.
- 3.Epstein M, Arakelian E, Tucker P, Dahlgren A. Managing sustainable working hours within participatory working time scheduling for nurses and assistant nurses: a qualitative interview study with managers and staffing assistants. J Nurs Manag. 2023;2023:1–12. 10.1155/2023/8096034. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Hong KJ, Cho SH, Jung EH. Analysis of the status of nursing shifts and nurses’ perceptions of work schedules in general wards based on the government guidelines for night shift work. J Korean Clin Nurs Res. 2021;27:165–78. 10.22650/JKCNR.2021.27.2.165. [Google Scholar]
- 5.Amindoust A, Asadpour M, Shirmohammadi S. A hybrid genetic algorithm for nurse scheduling problem considering the fatigue factor. J Healthc Eng. 2021;2021:5563651. 10.1155/2021/5563651. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Rerkjirattikal P, Huynh VN, Olapiriyakul S, Supnithi T. A goal programming approach to nurse scheduling with individual preference satisfaction. Math Probl Eng. 2020;2020:1–11. 10.1155/2020/2379091. [Google Scholar]
- 7.Karhula K, Turunen J, Hakola T, Ojajärvi A, Puttonen S, Ropponen A, et al. The effects of using participatory working time scheduling software on working hour characteristics and wellbeing: A quasi-experimental study of irregular shift work. Int J Nurs Stud. 2020;112:103696. 10.1016/j.ijnurstu.2020.103696. [DOI] [PubMed] [Google Scholar]
- 8.Koruca Hİ, Emek MS, Gulmez E. Development of a new personalized staff-scheduling method with a work-life balance perspective: case of a hospital. Ann Oper Res. 2023;1–28. 10.1007/s10479-023-05244-2. [DOI] [PMC free article] [PubMed]
- 9.Hainey CL. Implementation of a ward staff self-rostering system: improving morale and retention. Nurs Manag (Harrow). 2021;28:22–7. 10.7748/nm.2021.e1987. [DOI] [PubMed] [Google Scholar]
- 10.Huang YC, Hsieh YH, Hsia FY. A study on nurse day-off scheduling under the consideration of binary preference. J Ind Prod Eng. 2016;33:363–72. 10.1080/21681015.2015.1095805. [Google Scholar]
- 11.Ryu HJ, Jo YN, Lee WJ, Cheong JW, Moon BY, Ko YD. An effective staff scheduling for shift workers in social welfare facilities for the disabled. Algorithms. 2023;16:41. 10.3390/a16010041. [Google Scholar]
- 12.Shiri R, Karhula K, Turunen J, Koskinen A, Ropponen A, Ervasti J, et al. The effect of using participatory working time scheduling software on employee well-being and workability: a cohort study analysed as a pseudo-experiment. Healthc (Basel). 2021;9:1385. 10.3390/healthcare9101385. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Turunen J, Karhula K, Ropponen A, Koskinen A, Shiri R, Sallinen M, et al. The time-varying effect of participatory shift scheduling on working hour characteristics and sickness absence: evidence from a quasi-experiment in hospitals. Int J Environ Res Public Health. 2022;19:14654. 10.3390/ijerph192214654. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.O’Connell M, Barry J, Hartigan I, Cornally N, Saab MM. The impact of electronic and self-rostering systems on healthcare organisations and healthcare workers: A mixed‐method systematic review. J Clin Nurs. 2024;33:2374–87. [DOI] [PubMed] [Google Scholar]
- 15.Wynendaele H, Gemmel P, Pattyn E, Myny D, Trybou J. Systematic review: what is the impact of self-scheduling on the patient, nurse and organization? J Adv Nurs. 2021;77:47–82. 10.1111/jan.14579. [DOI] [PubMed] [Google Scholar]
- 16.Park KO, An JW, Han NK. A predictable pattern shift schedule model for improving nurses’ shift work, 42. Korea Institute for Health. and Social Affairs. 2022; pp. 258–76. 10.15709/hswr.2022.42.1.258.
- 17.Hamid M, Tavakkoli-Moghaddam R, Golpaygani F, Vahedi-Nouri B. A multi-objective model for a nurse scheduling problem by emphasizing human factors. Proc Inst Mech Eng H. 2020;234:179–99. 10.1177/0954411919889560. [DOI] [PubMed] [Google Scholar]
- 18.Benner P. From novice to expert. Am J Nurs. 1982;82:402–7. 10.2307/3462928. [PubMed] [Google Scholar]
- 19.Cho MS, Kwon IG, Kim KH, Cho YA. Revised clinical ladder system model based on nurses’ clinical competence and professional activities for nurses in general hospitals. J Korean Clin Nurs Res. 2018;24:324–35. 10.22650/JKCNR.2018.24.3.324. [Google Scholar]
- 20.Oh EY, Kim MW, Yoo HJ, Choi SH, Lee SR, Kim CS. Effects of nursing competence on job satisfaction across career stages. J Korean Acad Nurs Adm. 2023;29:258–67. 10.11111/jkana.2023.29.3.258. [Google Scholar]
- 21.O’Callahan K, Sitters S, Petersen M. You make the call’: improving radiology staff scheduling with AI-generated self-rostering in a medical imaging department. Radiography (Lond). 2024;30:862–8. 10.1016/j.radi.2024.03.014. [DOI] [PubMed] [Google Scholar]
- 22.Santamato V, Tricase C, Faccilongo N, Iacoviello M, Marengo A. Exploring the impact of artificial intelligence on healthcare management: a combined systematic review and machine-learning approach. Appl Sci. 2024;14:10144. 10.3390/app142210144. [Google Scholar]
- 23.Zhai Y, Li R, Yan Z. Research on application of meticulous nursing scheduling management based on data-driven intelligent optimization technology. Comp Intell Neurosci. 2022;2022:3293806. 10.1155/2022/3293806. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Leung F, Lau YC, Law M, Djeng SK. Artificial intelligence and end user tools to develop a nurse duty roster scheduling system. Int J Nurs Sci. 2022;9:373–7. 10.1016/j.ijnss.2022.06.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Faul F, Erdfelder E, Lang AG, Buchner A. G*Power 3: A flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behav Res Methods. 2007;39:175–91. 10.3758/BF03193146. [DOI] [PubMed] [Google Scholar]
- 26.Howard FM, Gao CA, Sankey C. Implementation of an automated scheduling tool improves schedule quality and resident satisfaction. PLoS ONE. 2020;15:e0236952. 10.1371/journal.pone.0236952. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.National Health Service Employers. Unsocial hours payments: find out about unsocial hours payments, including who [internet]hey apply [internet]o and [internet]he [internet]ates of pay [Internet]. London: National Health Service; 2022. https://www.nhsemployers.org/pay-pensions-and-reward/nhs-terms-and-conditions-of-service---agenda-for-change/nhs-terms-and-conditions-of-service-handbook/unsocial-hours-payments. [Google Scholar]
- 28.D’souza B, Rao SS, Muthana CG, Bhageerathy R, Apuri N, Chandrasekaran V, Prabhavathi D, Renukaradhya S. ROTA: A system for automated scheduling of nursing duties in a tertiary teaching hospital in South India. Health Inf J. 2021;27:3–12. 10.1177/14604582211001426. [DOI] [PubMed] [Google Scholar]
- 29.The Korean logitudinal survey of wmen and families. th wave questionnaires. 2022; p. 1 ~ 8. https://gsis.kwdi.re.kr/klowf/portal/data/dataInfoPage.do?pageSize=10%26pageIndex=1%26bbsId=SURVEY%26searchItem=all%26searchText=%26articleSeq=803.
- 30.Son DM, Jung YI. The effect of work-life balance on job satisfaction and turnover intention of hospital nurses: compared to female wage workers. Stress. 2019;27:268–76. 10.17547/kjsr.2019.27.3.268. [Google Scholar]
- 31.Murray M, Sundin D, Cope v. New. New graduate registered nurses’ knowledge of patient safety and practice: a literature review. J Clin Nurs. 2018;27:31–47. 10.1111/jocn.13785. [DOI] [PubMed] [Google Scholar]
- 32.American association of critical care nurses. AACN guiding principles for appropriate staffing. 2018. https://www.aacn.org/policy-and-advocacy/guiding-principles-for-staffing.
- 33.Griffiths P, Saville C, Ball J, Jones J, Pattison N, Monks T, et al. Nursing workload, nurse staffing methodologies and tools: a systematic scoping review and discussion. Int J Nurs Stud. 2020;103:103487. 10.1016/j.ijnurstu.2019.103487. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Shiffer D, Minonzio M, Dipaola F, Bertola M, Zamuner AR, Dalla Vecchia LA, et al. Effects of clockwise and Counterclockwise job shift work rotation on sleep and work-life balance on hospital nurses. Int J Environ Res Public Health. 2018;15:2038. 10.3390/ijerph15092038. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Karhula K, Puttonen S, Ropponen A, Koskinen A, Ojajärvi A, Kivimäki M, et al. Objective working hour characteristics and work-life conflict among hospital employees in the Finnish public sector study. Chronobiol Int. 2017;34:876–85. 10.1080/07420528.2017.1329206. [DOI] [PubMed] [Google Scholar]
- 36.Koh YR, Kim KJ, Cho HI, Song YR. Influence of shift work nurses’ unhealthy work scheduling on missed care and quality of care: mediating effect of fatigue. Korean Soc Nurs Res. 2024;8:83–98. 10.34089/jknr.2024.8.4.7. [Google Scholar]
- 37.Rizany I, Hariyati RTS, Afifah E, Rusdiyansyah. The impact of nurse scheduling management on nurses’ job satisfaction in army hospital: a cross-sectional research. Sage Open. 2019;9:2158244019856189. 10.1177/2158244019856189. [Google Scholar]
- 38.Pressley C, Garside J. Safeguarding the retention of nurses: a systematic review on determinants of nurse’s intentions to stay. Nurs Open. 2023;10:2842–58. 10.1002/nop2.1588. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The data that support the findings of this study are available from the corresponding author upon reasonable request. However, restrictions apply to the availability of these data, which contain sensitive institutional information and identifiable personnel data. Therefore, they are not publicly available. Data access may be granted upon request and with permission from Inha University Hospital.