The Effect of Nursing Intervention Model Using Mobile Nursing System on Pregnancy Outcome of Pregnant Women.

Due to the recent advancement in technology specifically mobile phones, these devices can be used in the hospital to monitor or speed up various activities, which are related to doctors and nurses. In literature, various mechanisms have been presented to resolve this issue, but none of these approaches have considered effectiveness of this technology in the development of a proper mobile nursing system, which is specifically designed for pregnant women. Therefore, in this paper, we have explored the effect of the intervention model based on the mobile nursing system on the pregnancy outcome of pregnant women. In this study, an Android-based mobile nursing monitoring system was adopted to monitor and transmit the human physiological data through physiological parameter monitoring equipment and continuously monitor the physiological parameter data of pregnant women. If the physiological health data of the pregnant woman was abnormal, it had to implement timely nursing intervention. In this study, 266 pregnant women in the electronic records (E-records) were selected as the research objects and divided into two groups according to the intervention method. Pregnant women in group A received routine physical examination during pregnancy, while those in group B received nursing intervention based on mobile nursing system. The results showed that the incidence of each indicator of pregnancy outcome in group B was significantly lower than that in group A, and the difference was statistically significant (P < 0.05). The nursing intervention model based on the mobile nursing system can effectively improve the pregnancy outcome. The mobile nursing system can help nursing staff find the abnormalities of pregnant women during pregnancy and give effective nursing measures in time, which helped improve the pregnancy outcomes, reduce the probability of adverse pregnancy outcomes, ensure the safety of puerperae and newborns, and lessen the delivery risk factors.

1. Introduction

There are many factors that cause bad pregnancy outcome, which are caused by the mutual influence of multiple factors such as parents, heredity, personal living habits, environment, and psychology [1]. With the gradual rise of social status, women have the possibility of receiving more education or work and are no longer confined to the family. Therefore, the marriage and childbearing ages of women have been postponed, and the elderly pregnant women have increased year by year. Many studies have proved that the rise of age leads to an increase in the incidence of pregnancy complications and complications, which greatly increases the risk of birth and the probability of adverse pregnancy outcomes, such as large babies and low birth weight babies [2]. With the development of modern society, people's living standards have improved, and there are unreasonable diets such as high-calorie and high-protein food intake. In addition, because pregnant women should not be too active during pregnancy, more and more pregnant women cannot control their own weight gain well. Studies have confirmed that there is a close relationship between the growth of pregnant women's own weight and pregnancy outcome. Obese people are more likely to develop gestational diabetes or hypertension during pregnancy, and the probability of cesarean section is greatly increased [3, 4]. With the rapid development of modern industrialized society, environmental pollution has become more and more serious, which has also led to the increasing probability of pregnant women's bad pregnancy outcomes, such as miscarriage and defects in newborns [5]. The occurrence of various bad pregnancy outcomes is accompanied by the risk of maternal and child death, which has had a huge impact on the public health and social problems of the quality of newborn populations in various countries, and it has brought huge economic and mental pressures to the family and society.

Placenta previa, postpartum hemorrhage, placental abruption, and premature rupture of membranes are all important factors that cause premature birth [6]. Studies have shown that advanced age, smoking, drug abuse, multiple pregnancy, curettage, and multiple abortions are all risk factors for placenta previa [7]. Puerperae, high blood pressure, smoking, drug use, and advanced age are also important reasons for placental abruption [8]. Lack of trace elements such as zinc in the body, lack of vitamin C, and reproductive tract infections can all lead to premature rupture of membranes [9]. In addition to the body's own factors such as soft birth canal injury and weak uterine contractions, multiple curettage, excessive births, overfrequency, poor physical fitness, and other chronic systemic diseases in pregnant women may all cause postpartum hemorrhage [10]. In order to reduce the occurrence of the abovementioned bad pregnancy outcomes, the Chinese government and local medical and health care departments have begun to pay attention to and strengthen the health care of pregnant women during pregnancy [11]. With the strengthening of publicity and the continuous improvement of health care during pregnancy, people have begun to pay more attention to health care during pregnancy, health awareness has gradually increased, and compliance has also improved. However, the incidence of adverse pregnancy outcomes such as newborns defects, low birth weight, pregnancy, or childbirth complications has not been found to be significantly reduced. Therefore, pregnancy nursing intervention has gradually been paid attention to by medical staff in order to reduce adverse pregnancy outcomes and improve population quality [12]. However, pregnant women usually go to the hospital for pregnancy checkups on a regular basis due to conditions or other influences. However, for pregnant women, especially pregnant women in the third trimester of pregnancy, it is necessary to detect vital signs in real time, so that they can be helped when the bodies are abnormal.

The technology and modern telemedicine involve many industries such as software technology, computer network, medical image processing, communication technology, and electronic information technology. The rapid development of hospital informatization has made the implementation of the mobile nursing system possible. The mobile nursing system covers the entire process from the patient entering the hospital, through the preadmission, hospitalization, and discharge. With the development of patient satisfaction as the core, it provides medical staff and managers with a process-based, informatized, paperless, automated, and intelligent one-stop nursing system management platform [13]. The mobile nursing system can record patient vital signs in real time and directly record and display the implementation of nursing-related diagnosis and nursing measures after admission, effectively reducing the occurrence of medical accidents. The computer automatically generates the electronic version of the nursing record, which makes the whole process of nursing diagnosis and treatment a qualitative change, allowing nurses to dispense, check, and execute more efficiently, and making the whole nursing process “paperless.” In addition, it can store the basic information of patients in time, realize the optimal medical process, minimize the medical error rate, and lessen the work intensity maximally. The electronic medical record can be moved and written, the medical decision-making is scientific and intelligent, and the quality assessment standard of the medical staff is digital.

Therefore, mobile nursing monitoring system based on Android this time is a way of health monitoring in remote mode. Physiological parameter monitoring equipment monitors and transmits human physiological data, real-time monitoring the physiological data such as heart rate, blood pressure, and body temperature of the user. If the physiological parameters of users are abnormal, an alarm can be issued in time. It was compared with pregnant women who have not undergone nursing intervention and only have routine physical examinations during pregnancy to explore the effect of this method on pregnancy outcome. In this paper, we have explored the effect of the intervention model based on the mobile nursing system on the pregnancy outcome of pregnant women. In this study, an Android-based mobile nursing monitoring system was adopted to monitor and transmit the human physiological data through physiological parameter monitoring equipment and continuously monitor the physiological parameter data of pregnant women. If the physiological health data of the pregnant woman was abnormal, it had to implement timely nursing intervention. In this study, 266 pregnant women in the electronic records (E-records) were selected as the research objects and divided into two groups according to the intervention method. Pregnant women in group A received routine physical examination during pregnancy, while those in group B received nursing intervention based on mobile nursing system.

The remaining sections of the paper are managed and represented accordingly as given below.

In subsequent section, the android based mobile nursing monitoring system is described and presented in detailed along with its various portions or sections. Experimental results and observations were reported in Section 3 of the manuscript. These results have verified the exceptional performance of the android based mobile nursing system. A detailed discussion section is provided, which is followed by comprehensive and brief concluding remarks sections. Finally, reference materials are provided.

2. Android Based Mobile Nursing Management System

2.1. Research Objects

In this study, the clinical data of 266 pregnant women were extracted from the electronic record (E-record) system. When the data were extracted, it was necessary to exclude I. pregnant women who were selected to induce labor or abortion due to family planning restrictions, personal or family factors after pregnancy; II. pregnant women who suffered from mental illness who did not know their prepregnancy healthcare or had poor compliance and were unwilling to cooperate with the investigation. According to the treatment method of the patients, they were divided into two groups: pregnant women in group A received routine physical examination during pregnancy, while those in group B received nursing intervention based on mobile nursing system.

2.2. Mobile Nursing System

In this study, a remote nursing monitoring system based on the Android platform was adopted, which was functioned with remote guidance monitoring and physiological data monitoring to provide pregnant women with medical interaction, realize seamless connection and all-round medical monitoring services between hospitals and families, and help nurses give timely nursing interventions (Figure 1).

Figure 1

The remote nursing monitoring system based on the android platform.

In the beginning, the heart rate, blood pressure, and body temperature of user were detected by the remote medical monitoring system through various physiological data collection terminals. After that, the collected physiological data were performed with D/B conversion, and the wireless WiFi method was used to communicate with the Android mobile phone monitoring terminal (MPMT). The Android MPMT processed the detected physiological data such as body temperature and blood pressure and then monitored and stored these data. In addition, the Android MPMT transmitted the acquired data to the service system of the remote monitoring center through interaction with the wire beam electrode (WBE) service platform of the remote monitoring center, so that the remote monitoring center can monitor these data in real time. Users, their relatives, and medical staff can log in at any time and view the general trend of current or past physiological data and various other related service information.

2.3. Related Algorithms for Physiological Parameter Collection

Because the monitoring system remotely detects the physiological data of the user, unlike a hospital that can help patients solve problems at any time, it is extremely necessary to predict the physiological state of the patient in real time. For example, if the heart abnormality cannot be found in time for heart disease patients, it may lead to sudden death. If the system can predict the physical condition of future patients in time, it can give nurses enough time to prepare solutions so as not to overwhelm nurses when problems arise. The prediction system not only reduces the risk of patients, but also reduces the pressure on nurses and effectively improves the nursing effect. Therefore, a time series prediction algorithm was applied in this study to predict the blood pressure, electrocardiograph (ECG), and other conditions of patients.

At this stage, time series prediction methods roughly include two major categories: statistical methods and artificial intelligence methods [14, 15]. In addition, it can be divided into global model and local model according to the difference of modeling principle and structure [16] (Figure 2). Statistical methods can be divided into two methods: qualitative and quantitative prediction [17]. The qualitative prediction method is based on the existing relevant theories and knowledge and obtains the prediction results based on intuition. Therefore, the accuracy of its conclusions is mostly determined by the subjective thoughts of the judges, so the accuracy of the prediction results is poor. Quantitative prediction is based on existing experimental data to establish and calculate a certain mathematical model to obtain prediction data. Compared with qualitative prediction, the accuracy of prediction results has been effectively improved, so it has also attracted more attention.

Figure 2

Comparison of prediction principles between global model and local model. (a) Global model. (b) Local model.

2.3.1. Global Model Prediction

Statistical methods represented by autoregressive and moving average model (ARIMA) achieved the time series fitting by determining model parameters and model coefficients. Then, the fitted model was adopted to predict the future trend, which showed good prediction effect on linear systems [18]. The specific statistical method was given as follows:

(1) Autoregressive (AR) Model. AR model is a dynamic description of random process with good stability and is suitable for the prediction of linear time series data. In this model, the current time value y1 of the time series was represented by the linear combination of the p historical time values of the sequence plus the white noise perturbation a1, as shown in the following equation:

The above equation represented the p-order AR model AR (p), a1 represented the white noise disturbance, showing the sequence when the mean was equal to 0 and the variance was σ2, and ∅ (i = 1, 2,…, p) was the model coefficient. The definition of AR model can explain that its function was to return the historical data of the sequence itself.

(2) Moving Average (MA) Model. In MA model, the linear combination of the white noise disturbance term in the past moment and the white noise disturbance term at the current moment were used to represent the current moment values of the time series, as shown in the following equation:

The above equation represented the q-order MA model MA (q), where the weighting factor ∅ (i = 1, 2,…, q) was the model coefficient. The MA model can run in reverse, and it can also be used to predict linear system time series.

(3) Autoregressive Moving Average (ARMA) Model. ARMA(p, q) can be expressed as the following equation:

As shown in the above equation, the ARMA model can be classified into two types: AR model and MA model. It is generally used to predict linear and stable time series modeling. Because modeling is easy to understand and easy to operate, it is widely used in many fields. However, in reality, most of the environment is not stable, or the sequence data is nonlinear, and ARMA is a linear model, so it cannot meet the requirements of such data modeling accuracy.

(4) Autoregressive and Moving Average (ARIMA) Model. The structural model of ARIMA(p, d, q) can be expressed as the following equation:

In the above equation, ∇ = 1 − B represented the difference operator, and the nonstationary time series y1 can be obtained by using the d-order difference processing to obtain the stationary time series ∇y1, ∅(B)=1 − ∅1B − ⋯−∅B, and ∅(B)=1 − ∅1B − ⋯−∅B. It can be explained from the above equation that when d = 0, the ARIMA (p, d, q) model was the ARIMA (p, q) model. The process of the ARIMA time series prediction was shown in Figure 3.

Figure 3

The process of the ARIMA time series prediction.

The ARIMA model further improved the ARMA model, applying difference to deal with non-stationary data. Although there were applications within a certain range, it had a strong pertinence to application data and cannot adapt to some complex nonlinear data.

With the continuous advancement of modern machine learning technology, some artificial intelligence methods have gradually emerged and are widely used in various fields of predicting time series [19]. Compared with traditional statistical methods, these artificial intelligence methods have obvious advantages in processing time series data with complex nonlinearities, and some data can build the mapping relationship between input and output without human intervention. These methods of predicting sequences have been widely used at this stage. The least squares support vector machine (LS-SVM) method was briefly described in this study.

LS-SVM is mostly used to deal with function estimation or nonlinear regression. The initial weight space of the regression model was shown in the following equation:

In the equation above, ω represented the weight vector, b was the bias term, and ∅ (x) was the mapping function from the input space to the high-dimensional feature space. The optimization steps of LS-SVM in the initial space can be as shown in the following equation:

The constraints were defined as follows:

In equation (7), y is the output value, γ is the regularization parameter, which is the penalty for the error term, and e is the regression error of the k-th training sample.

Applying Lagrangian function to solve the above optimization, the following equation can be obtained:

It could solve all the variables in equation (8), where a is a Lagrangian multiplier, and the inequality constraint term is transformed into an equality constraint term according to KKT:

After training to obtain the values of ω and e, the following linear system can be obtained:

In equation (10), (1=1,1,…,1) was a matrix in N × 1, (a=a1, a2,…,a), and K referred to the kernel function.

Therefore, the LS-SVM prediction model can be obtained as follows:

In equation (11), represented the output value of the prediction model, (x, x=∅(x)∅(x)), and the following radial basis kernel function was used: K(x, x)=e|.

2.4. Local Model Prediction

The main purpose of the local model is to find the nearest neighbor set of the query sample. Its focus is to select the similarity measurement algorithm. The appropriate measurement algorithm can ensure that the local model has a better prediction effect. In this study, the multitask LA-SVM algorithm and the K-means algorithm method were adopted and combined with semisupervised and supervised learning algorithms. The input values were clustered through the K-mean algorithm, adjacent samples were classified into the same cluster, and each of them was trained and simulated by the MTLS-SVM algorithm to obtain the predicted output value. This method could not only make up for the shortcomings of mining sequence information in traditional algorithms, but also did not consider the poor prediction accuracy caused by comprehensive intersequence correlation and other reasons. At the same time, it reduces the dimensionality in the support vector machine and effectively improves the operating efficiency.

The algorithm mainly included three parts: data preprocessing, task construction, and model training (Figure 4):

Figure 4

The time series prediction models based on K-means and MTLS-SVM.

The data preprocessing was described as follows.

It was supposed that the time series {y1, y2,…, y, y,…, y} was sampled at the same time interval, t was the current time, and the predicted time series was to estimate the time t + h based on the historical sampling value. The y was shown in the following equation:

In equation (12), h represented the prediction step length; when h = 1, it meant single-step prediction, and when h > 1, it meant multistep prediction. f represented the prediction model, and n represented the delay time. The collected observation value {y1, y2,…, y, y,…, y} was clustered to construct a data set sample through the K-means algorithm.

For the task construction (shown in Figures 4 and 5), the single task and multiple tasks constructed in the time series, the combination of multitask learning methods and time series prediction focuses mainly on the close relationship between adjacent time points. In the past, single task was used. The learning method was to predict the value at time t + h (h ≥ 1) from the observation value at n times, and the model output value was single. The multitask learning method can jointly represent the prediction output of t + 1, t + 2,…, t + k, which had the effect of inductive bias and improves the accuracy of prediction.

Figure 5

Multitask construction in time series.

Applying the multitask learning method only needed to build a multioutput prediction model, and the output values of k tasks could be obtained at a time. According to the time series prediction results, the appropriate k tasks were selected, and a corresponding training set was constructed for each task.

The model training was described as follows.

It could train the corresponding training sets of k learning tasks and the MTLS-SVM model synchronously. Because these tasks were to predict similar time values synchronously, there was a parallel relationship and a close relationship among them. Throughout the training process, tasks can be restricted to each other to achieve the effect of inductive bias, thereby improving the model's ability to predict unknown data (Algorithm 1).

Algorithm 1

The process of predicting the time series based on MTLS-SVM was clarified as follows.

After the required estimated value was selected, it can be compared with the measured data to illustrate the accuracy of the algorithm prediction in this study.

2.5. Statistical Methods

The SPSS22.0 was adopted to perform statistical analysis on the data, and descriptive statistics were performed on the pregnant women pregnancy outcome of each group. The data between the two groups were compared using analysis of variance, and P < 0.05 was considered statistically significant.

3. Results and Evaluation

In this section, a detailed discussion on various evaluation metrics and performance of the proposed system is presented.

3.1. Algorithm Model Experiment Results

In order to verify the practicability of the research algorithm and obtain accurate experimental results, the case data of 95 patients diagnosed with heart disease and hypertension were extracted. After data collection and processing, K-means algorithm was used to implement the clustering algorithm. In this study, the appropriate value of k was determined firstly so that the results that can be easily obtained showed higher accuracy. The results when the k value was 2, 3 and 4, 5 were simulated through different subsets, as shown in Figures 6–9.

Figure 6

Accuracy test results when K = 2.

Figure 7

Accuracy test results when K = 3.

Figure 8

Accuracy test results when K = 4.

Figure 9

Accuracy test results when K = 5.

The number of sample values contained in data sets 1, 2, and 3 was 198, 427, and 562, respectively. It can be explained that when the data set sample was small, the preaccuracy was best when k = 2; and when the data set sample was large, the prediction accuracy was the best when the k value was 3.

3.2. Comparison on Pregnancy Outcome Results

The comparison results of pregnancy outcome indicators of the two groups are shown in Figure 10. The incidence of related pregnancy outcome indicators of group A was significantly higher than that of group B, and the difference was statistically significant (P < 0.05).

Figure 10

Comparison on pregnant women pregnancy outcome (suggested P < 0.05 compared with the control group).

4. Discussion

Poor pregnancy outcome is the result of a variety of factors before and during pregnancy, which seriously threatens the safety of mothers and babies. Some pregnant women have poor prognostic effects, which bring huge pressure to the family and society. Poor pregnancy outcome includes various complications during childbirth, various complications during pregnancy, induction of labor, and premature birth caused by various pathological factors during pregnancy [20]. Some studies all over the world have proved that nursing intervention can effectively reduce the probability of adverse pregnancy outcome, and there is an effective method to ensure the health of pregnant women during pregnancy and after delivery [12, 21]. However, traditional nursing interventions cannot guarantee that pregnant women are always concerned about their physical conditions. Therefore, this study was made to explore the impacts of the nursing intervention model based on the mobile nursing system on the pregnancy outcome, provide strong data support for further improving the health care effect during pregnancy, and reduce the occurrence of undesirable pregnancy outcomes.

In this article, the K-means algorithm and the multitask LS-SVM algorithm were combined, and the human physiological parameter monitoring system was combined; the K-means algorithm was adopted to cluster the data, and then the model was trained through the multitask LS-SVM algorithm; and more accurate results were obtained in the prediction of physiological parameters of patients such as ECG and blood pressure. In addition, the pregnancy outcomes were compared between pregnant women who had not undergone special nursing interventions and pregnant women who had been given nursing interventions based on the mobile nursing system. It can be found that, regarding the pregnancy outcome indicators of placenta previa, premature delivery, fetal distress, postpartum hemorrhage, premature rupture of membranes, urinary tract infection, polyhydramnios, and cesarean section, the probability of occurrence of various adverse pregnancy outcomes in group B was significantly lower in contrast to that in group A, and the difference was obvious with statistical significance (P < 0.05). This shows that the nursing intervention model based on the mobile nursing system can effectively improve the pregnancy outcome. Preterm birth refers to babies whose gestation period is 28 weeks but less than 37 weeks at the time of delivery. The research results show that preterm birth accounts for 8.4% of the total number of bad pregnancy outcomes in puerperae, which is roughly the same as the average incidence of 5%∼15% in China previously reported. Postpartum hemorrhage means that pregnant women lose more than 500 mL of blood within 24 hours after delivery, and the prevalence rate accounts for 28%–3% of the total number of deliveries. The results of the study showed that postpartum hemorrhage accounted for 9.4%. This situation may be related to the increase in the cesarean section rate. In this study, the cesarean section rate was as high as 54.1%.

5. Conclusion

In this paper, we have explored the effect of the intervention model based on the mobile nursing system on the pregnancy outcome of pregnant women. In this study, an Android-based mobile nursing monitoring system was adopted to monitor and transmit the human physiological data through physiological parameter monitoring equipment and continuously monitor the physiological parameter data of pregnant women. If the physiological health data of the pregnant woman was abnormal, it had to implement timely nursing intervention. In this study, 266 pregnant women in the electronic records (E-records) were selected as the research objects and divided into two groups according to the intervention method. Pregnant women in group A received routine physical examination during pregnancy, while those in group B received nursing intervention based on mobile nursing system. In summary, this study made up for the shortcomings of inability to monitor in real time and effectively respond to the emergencies of pregnant women. During pregnancy, there would be various factors that will affect the pregnancy outcome. Only by doing a good job in pregnancy health care, timely discovering pregnancy risk factors and giving corresponding nursing interventions, can it effectively reduce the appearance of bad pregnancy outcomes, ensure the health of pregnant women and newborns, and improve the birth quality of the population.