Assessment of the value of complication risk prediction models following interventional therapy for hepatocellular carcinoma in nursing decision-making

Abstract

BACKGROUND

Transcatheter arterial chemoembolization (TACE) is a primary interventional modality for intermediate-to-advanced hepatocellular carcinoma (HCC). However, the high incidence of post-procedural complications significantly compromises patient recovery and prognosis. Currently, clinical nursing assessment lacks a multifaceted risk prediction tool that integrates multiple risk factors.

AIM

To establish a predictive model for postoperative complications in patients with HCC undergoing TACE.

METHODS

A retrospective analysis was conducted on 386 patients with HCC who underwent interventional therapy at our hospital from January 2023 to December 2024. Patients were divided into a complication group (n = 104) and a control group (n = 282) based on postoperative complication occurrence. General clinical data and nursing-related indicators were collected and compared between groups. Multivariate logistic regression analysis was used to identify independent risk factors, construct a risk prediction model, and validate its discriminatory power and calibration by using receiver operating characteristic curve and Hosmer-Lemeshow test.

RESULTS

The incidence of complications following interventional therapy for HCC in this study was 26.94%. Factors associated with complications included age ≥ 60 years, liver cirrhosis, vascular invasion, procedure duration ≥ 2 hours, TACE sessions > 2, Child-Pugh grade B, tumor diameter ≥ 5 cm, nutritional risk (Nutritional Risk Screening 2002 ≥ 3), anxiety (Self-Rating Anxiety Scale ≥ 50), depression (Self-Rating Depression Scale ≥ 53), caregiver burden (high Zarit Burden Interview Score), functional independence (low Exercise of Self-Care Agency Scale Score), quality of life (low Quality of Life Instruments for Cancer Patients - General Module Score), and low compliance with early postoperative activity and exercise were all independent risk factors (P < 0.05). The predictive model demonstrated a C-index of 0.773, an area under the curve of 0.936, sensitivity of 93.25%, specificity of 84.96%, and good calibration (Hosmer-Lemeshow test, P = 0.382).

CONCLUSION

The TACE postoperative complication prediction model derived during the research combines multidimensional clinical and nursing predictors, which proves a high predictive quality and clinical usefulness. It provides healthcare professionals with a scientifically grounded assessment tool to facilitate risk stratification management and precision nursing.

Key Words: Hepatocellular carcinoma; Interventional therapy; Transcatheter arterial chemoembolization; Complication risk; Risk prediction model; Nursing decision-making

Core Tip: This study developed and validated a multidimensional integrated model for predicting the risk of complications following transcatheter arterial chemoembolization in patients with hepatocellular carcinoma. By innovatively incorporating clinical characteristics and procedural parameters with nursing-specific indicators including psychological status, nutrition, and self-care capacity, the model achieved high predictive accuracy (area under the curve = 0.936). The model offers nurses a logical and user-friendly decision-support model that helps them to predict complications early and accurately distribute nursing resources.

INTRODUCTION

Hepatocellular carcinoma (HCC) is a high incidence and mortality malignant tumor worldwide[1]. One of the most effective treatments in the management of the progressive stages of HCC is transcatheter arterial chemoembolization (TACE) which is effective in controlling tumor progression though accompanied by high rates of postoperative complications that largely influence patient prognosis[2-4]. Today, nursing interventions are mostly based on experience and standard guidelines, without risk prediction tools of integrated multifactorial analysis, which is an obstacle to precision nursing management implementation[5,6]. Thus, it is important to create a model of complication risks prediction that incorporates both clinical features and nursing-related factors, ensuring that the nursing decision-making in TACE patients can be early-warned and supported. The purpose of this study was to combine the multidimensional risk factors using logistic regression to develop and test a model of risk prediction of postoperative complications using TACE. The aim of this approach is to equip nursing staff with a scientific and practical risk assessment instrument, to streamline the allocation of the nursing resources, and to improve the nursing quality and patient outcomes.

MATERIALS AND METHODS

General information

In this research, 386 cases of HCC were participants, and the cases were chosen January 2023-December 2024. Patients were divided into a complication group (n = 104) and a control group (no complications, n = 282) based on whether complications occurred after interventional therapy. Detailed characteristics of both groups were presented in Table 1. Comparisons of clinical data including gender, body mass index (BMI), underlying medical conditions (hypertension, diabetes, hyperlipidemia), and history of alcohol consumption and smoking revealed no statistically significant differences (P > 0.05). This study was approved by the institutional ethics committee. Complications following interventional therapy included upper abdominal pain, fever, nausea and vomiting, intra-abdominal hemorrhage, intra-abdominal infection, liver abscess, and tachycardia.

Table 1 Distribution of postoperative complications in hepatocellular carcinoma patients following transcatheter arterial chemoembolization (n = 104).

Complication types	n	Percentage (%)
Nausea and vomiting	42	40.38
Upper abdominal pain	12	11.54
Fever	29	27.88
Liver abscess	6	5.77
Intra-abdominal hemorrhage	3	2.88
Tachycardia	3	2.88
Intra-abdominal infection	9	8.65

This retrospective cohort study enrolled patients who underwent their first TACE treatment between January 2023 and December 2024. The follow-up concluded on June 30, 2025, ensuring a minimum follow-up duration of six months for all participants. Complications defined in this study included both early (≤ 30 days) and delayed (> 30 days) post-procedural events, such as liver abscess, intra-abdominal infection, hemorrhage, and hepatic failure. Data were comprehensively collected and verified through outpatient reviews, readmission records, or telephonic follow-ups.

Inclusion criteria: Patients were eligible for inclusion if they: (1) Met the diagnostic criteria for primary HCC[7], as confirmed by clinical presentation, imaging (computed tomography/magnetic resonance imaging), and pathological examination; (2) Fulfilled the indications for interventional therapy according to the NCCN Clinical Practice Guidelines in Oncology[8] and elected to undergo the procedure; (3) Possessed complete clinical records and had no contraindications for interventional maneuvers; and (4) Provided informed consent from both the patients and their legal guardians for study participation.

Exclusion criteria: (1) Presence of severe systemic comorbidities such as major cardiovascular or cerebrovascular disease, severe neurological disorders, or respiratory failure; (2) Receipt of immunosuppressive agents, glucocorticoids, antibiotics, radiotherapy, or chemotherapy within one month prior to admission; (3) Severe autoimmune diseases (e.g., antibody-mediated immunodeficiency disorders, immunodeficiency syndromes, dysimmunological disorders); (4) Child-Pugh grade C liver function[9]; (5) Coexisting coagulation disorders; and (6) Coexisting other malignancies or metastatic HCC.

Methods

The electronic medical record system of the hospital was searched to extract clinical data on a retrospective basis. The period of follow-up was over on June 30, 2025, with all patients having at least six months of observation. Complications were considered as any procedure related negative events that occurred since the commencement of TACE to the follow up. These events were classified as early [i.e., symptoms of post-embolization syndrome (pain, pyrexia, emesis, etc.)] and delayed (i.e., liver abscess, abdominal infection and bleeding). All complications were adjudicated based on clinical diagnosis or confirmed via objective imaging findings.

Data collection and cross-verification were independently performed by two uniformly trained researchers, covering the following dimensions.

General information: (1) Demographic characteristics (gender, age, BMI); (2) History of underlying diseases (diabetes, hypertension, hyperlipidemia, and cirrhosis, etc.); (3) Lifestyle habits [smoking history (defined as ≥ 1 cigarette per day for ≥ 6 months), alcohol consumption history (defined as ≥ 1 episode per week with single intake > 30 mL, sustained for ≥ 6 months)]; and (4) TACE procedure parameters (vascular invasion status, maximum tumor diameter, number of tumors, preoperative Child-Pugh grade, number of TACE treatments, lymph node metastasis status).

The nursing-related indicators and assessment methods: (1) Self-care ability at 72 hours post-surgery: Assessed using the Exercise of Self-Care Agency Scale (ESCA). This 43-item scale covers four dimensions: Sense of responsibility for self-care, self-care skills, health knowledge level, and self-concept. The total score ranges from 0 point to 172 points, with higher scores indicating stronger self-care ability. Assessment occurs within 72 hours post-surgery; (2) Quality of life assessment: Evaluated using the Quality-of-Life Instruments for Cancer Patients - General Module (QLICP-LI) with liver cancer-specific items. This scale consists of 30 questions, which assess four domains: Physical functioning, psychological functioning, social functioning and symptoms/side effects. The total scores will be between 0 and 100, with a high score being a sign of a better quality of life. Time of assessment: 1 week after surgery; (3) Primary caregiver burden assessment: Zarit Burden Interview (ZBI). This is a 22-item scale that encompasses the personal burden, emotional burden, and social burden. The total scores are 0-88, above 40 indicates high burden. Primary caregivers completed the assessment within 72 hours postoperatively; (4) Psychological state assessment: Evaluated using the Self-Rating Anxiety Scale (SAS) and Self-Rating Depression Scale (SDS). The positive anxiety and depression states, respectively, are a total score 50 or above on SAS and 53 or above on SDS. Time of assessment: 24 hours before surgery; (5) Nutritional Risk Screening (NRS): Evaluated with the NRS 2002 scale. A total score ≥ 3 indicates nutritional risk. Evaluation during 24 hours preoperative; and (6) Postoperative early mobility and functional exercise adherence: Document patient adherence to bedside mobility and functional exercises during 24-72 hours after surgery. Good compliance was defined as 80% of the prescribed volume and poor compliance as below 80% of the prescribed volume.

Statistical analysis

This study involved continuous variables, which were tested and observed to have a normal distribution. Independent samples t-tests were used to analyze quantitative data and 2-sample tests were used to analyze categorical data with the help of SPSS 24.0 software. The independent risk factors in complications after HCC interventional therapy were determined using multivariate logistic regression analysis and a risk prediction model was developed using the regression coefficients. The rms package of the R 3.6.1 software was used to model visualization. Calibration of the model was done through the Hosmer-Lemeshow test and calibration curves, and the calibration performance was measured through the calculation of the area under the curve (AUC) using the receiver operating characteristic curve. The significance threshold was set at P < 0.05.

RESULTS

Comparison of clinical data between two groups

This study included 386 patients undergoing interventional therapy for HCC, among whom 104 developed complications, yielding an incidence rate of 26.94% (Table 1). The complication group and control group showed statistically significant differences in age, cirrhosis, vascular invasion, operative time, number of TACE procedures, Child-Pugh classification, tumor diameter, and all nursing-related indicators (including NRS 2002 Score, SAS Score, SDS Score, ZBI Score, QLICP-LI Score, and compliance with early postoperative activity and functional exercise) (P < 0.05). No statistically significant differences were observed between the two groups in terms of gender, BMI, history of underlying diseases (hypertension, diabetes, hyperlipidemia), smoking history, drinking history, number of tumors, or lymph node metastasis (P > 0.05), as shown in Table 2.

Table 2 Comparison of clinical characteristics between the two groups of patients, n (%)/mean ± SD.

Items	Complication group (n = 104)	Control group (n = 282)	t/χ2	P value
Gender			0.166	0.684
Male	54 (51.92)	153 (54.26)
Female	50 (48.08)	129 (45.74)
Age (years)			16.100	< 0.001
≥ 60	74 (71.15)	136 (48.23)
< 60	30 (28.85)	146 (51.77)
Body mass index (kg/m2)			2.436	0.119
≥ 23.9	71 (68.27)	168 (59.57)
< 23.9	33 (31.73)	114 (40.43)
Hypertension			1.314	0.252
Yes	23 (22.12)	48 (17.02)
No	81 (77.88)	234 (82.98)
Diabetes			0.033	0.855
Yes	42 (40.38)	111 (39.36)
No	62 (59.62)	171 (60.64)
Hyperlipidemia			1.097	0.295
Yes	25 (24.04)	83 (29.43)
No	79 (75.96)	199 (70.57)
Liver cirrhosis			37.083	< 0.001
Yes	78 (75.00)	113 (40.07)
No	26 (25.00)	169 (59.93)
Smoking history			0.548	0.459
Yes	40 (38.46)	97 (34.40)
No	64 (61.54)	185 (65.60)
Alcohol consumption history			0.352	0.553
Yes	35 (33.65)	86 (30.50)
No	69 (66.35)	196 (69.50)
Tumor diameter (cm)			24.961	< 0.001
≥ 5	66 (63.46)	99 (35.11)
< 5	38 (36.54)	183 (64.89)
Number of tumors			0.275	0.600
Single discharge	43 (41.35)	125 (44.33)
Multiple discharges	61 (58.65)	157 (55.67)
Vascular invasion			35.225	< 0.001
Yes	80 (76.92)	121 (42.91)
No	24 (23.08)	161 (57.09)
Operative duration (hours)			14.712	< 0.001
≥ 2	57 (54.81)	94 (33.33)
< 2	47 (45.19)	188 (66.67)
Lymph node metastasis			1.975	0.160
Yes	59 (56.73)	182 (64.54)
No	45 (43.27)	100 (35.46)
Child-Pugh grade			15.524	< 0.001
Grade A	57 (54.81)	213 (75.53)
Grade B	47 (45.19)	69 (24.47)
Transcatheter arterial chemoembolization procedures (times)			14.243	< 0.001
> 2	85 (81.73)	173 (61.35)
≤ 2	19 (18.27)	109 (38.65)
Preoperative Nutritional Risk Screening 2002 Score			6.919	0.009
≥ 3	68 (65.38)	142 (50.35)
< 3	36 (34.62)	140 (49.65)
Postoperative Exercise of Self-Care Agency Scale Score	87.32 ± 12.15	102.46 ± 10.83	11.784	< 0.001
Postoperative Quality of Life Instruments for Cancer Patients - General Module Score	52.17 ± 8.26	68.43 ± 7.95	17.641	< 0.001
Self-Rating Anxiety Scale Score			11.026	0.001
≥ 50	49 (47.12)	82 (29.08)
< 50	55 (52.88)	200 (70.92)
Self-Rating Depression Scale Score			10.486	0.001
≥ 53	51 (49.04)	88 (31.21)
< 53	53 (50.96)	194 (68.79)
Zarit Burden Interview Score	45.63 ± 9.27	32.15 ± 8.46	13.530	< 0.001
Early postoperative ambulation (%)			25.165	< 0.001
≥ 80	28 (26.92)	157 (55.67)
< 80	76 (73.08)	125 (44.33)
Functional exercise completion rate (%)			24.741	< 0.001
≥ 80	22 (21.15)	139 (49.29)
< 80	82 (78.85)	143 (50.71)

Multifactorial analysis of complications following interventional treatment for HCC

Using complications after interventional treatment for HCC as the dependent variable (yes = 1, no = 0), statistically significant factors from Table 2 (e.g., age, tumor diameter, operative duration, TACE procedures) were incorporated as independent variables into a logistic regression model. Results indicated that age, liver cirrhosis, vascular invasion, operative duration, tumor diameter, TACE procedures, Child-Pugh B grade, preoperative NRS 2002 Score, postoperative ESCA Score, QLICP-LI Score, SAS Score, SDS Score, ZBI Score, early postoperative ambulation compliance, and functional exercise completion rate were all independent risk factors for complications following interventional therapy for HCC (P < 0.05), as shown in Tables 3 and 4.

Table 3 Variable assignment table.

Related factors	Variable names	Variable assignment
Age	X1	≥ 60 years = 1, < 60 years = 0
Liver cirrhosis	X2	Yes = 1, no = 0
Tumor diameter	X3	≥ 5 cm = 1, < 5 cm = 0
Vascular invasion	X4	Yes = 1, no = 0
Operative duration	X5	≥ 2 hours = 1, < 2 hours =0
Child-Pugh grade	X6	Grade B = 1, grade A = 0
Transcatheter arterial chemoembolization Procedures	X7	> 2 times = 1, ≤ 2 times = 0
Preoperative Nutritional Risk Screening 2002 Score	X8	≥ 3 scores = 1, < 3 scores = 0
Postoperative Exercise of Self-Care Agency Scale Score	X9	Actual value
Postoperative Quality of Life Instruments for Cancer Patients-General Module Score	X10	Actual value
Self-Rating Anxiety Scale Score	X11	≥ 50 scores = 1, < 50 scores = 0
Self-Rating Depression Scale Score	X12	≥ 53 scores = 1, < 53 scores = 0
Zarit Burden Interview Score	X13	Actual value
Early postoperative ambulation compliance	X14	< 80% = 1, ≥ 80% = 0
Functional exercise completion rate	X15	< 80% = 1, ≥ 80% = 0

Table 4 Results of logistic regression analysis for factors influencing the risk of complications following interventional treatment for hepatocellular carcinoma.

Variable names	Β	SE	χ2	P value	Odds ratio (95%CI)
Constant	-2.867	0.218	-	-	-
Age	0.325	0.134	5.882	0.015	1.384 (1.064-1.800)
Liver cirrhosis	0.481	0.211	5.197	0.023	1.618 (1.070-2.446)
Tumor diameter	0.296	0.108	7.512	0.006	1.344 (1.088-1.661)
Vascular invasion	1.114	0.369	9.114	0.003	3.047 (1.478-6.279)
Operative duration	0.563	0.271	4.316	0.038	1.756 (1.032-2.987)
Child-Pugh grade	0.601	0.254	5.599	0.018	1.824 (1.109-3.001)
Transcatheter arterial chemoembolization Procedures	0.742	0.302	6.037	0.014	2.100 (1.162-3.796)
Preoperative Nutritional Risk Screening 2002 Score	0.309	0.114	7.347	0.007	1.362 (1.089-1.703)
Postoperative Exercise of Self-Care Agency Scale Score	0.517	0.153	11.418	0.001	1.677 (1.242-2.263)
Postoperative Quality of Life Instruments for Cancer Patients - General Module Score	0.448	0.203	4.870	0.027	1.565 (1.051-2.330)
Self-Rating Anxiety Scale Score	0.635	0.247	6.609	0.010	1.887 (1.163-3.062)
Self-Rating Depression Scale Score	0.557	0.263	4.485	0.034	1.745 (1.042-2.923)
Zarit Burden Interview Score	0.751	0.165	20.716	< 0.001	2.119 (1.534-2.928)
Early postoperative ambulation compliance	0.495	0.223	4.927	0.026	1.640 (1.060-2.540)
Functional exercise completion rate	1.011	0.113	80.047	< 0.001	2.748 (2.202-3.430)

Construction and evaluation of the predictive model

This study employed a logistic regression method to construct a predictive model for postoperative complications following HCC intervention. The regression model expression was: Logit(P) = -2.867 + 0.325 × age + 0.481 × liver cirrhosis + 0.296 × tumor diameter + 1.114 × vascular invasion + 0.563 × operative duration + 0.601 × Child-Pugh grade + 0.742 × TACE procedures + 0.309 × preoperative NRS 2002 Score + 0.517 × postoperative ESCA Score + 0.448 × postoperative QLICP-LI Score + 0.635 × SAS Score + 0.557 × SDS Score + 0.751 × ZBI Score + 0.495 × early postoperative ambulation compliance + 1.011 × functional exercise completion rate. After validation, this predictive model demonstrated excellent clinical discriminatory ability with a consistency index of 0.773 (95% confidence interval: 0.746-0.853), indicating reliable predictive efficacy.

Diagnostic value analysis of the predictive model

The diagnostic value of the risk prediction model was assessed using receiver operating characteristic curve analysis. Results showed that the model achieved an AUC of 0.936 (0.884-0.988), with sensitivity and specificity of 93.25% and 84.96%, respectively (Figure 1). The Hosmer-Lemeshow test and calibration curves were used to evaluate the model calibration. The χ² statistic was χ² = 7.354 (P = 0.382), indicating good model calibration. The calibration curve had high fit with the ideal curve, which further confirms the predictive power of the model. Moreover, the analysis of decision curves revealed that the predictive model developed in this study has higher net benefit than other traditional prediction models in a wide range of threshold probabilities, as Figure 2 reveals.

Figure 1 Receiver operating characteristic curve of the risk prediction model for predicting the risk of complications following interventional treatment for hepatocellular carcinoma. AUC: Area under the curve; FPR: False positive rate; TPR: Talse positive rate.

Figure 2  Calibration curve.

DISCUSSION

Research has shown[10-12] that more than 80% of patients with HCC miss the chance of a curative operation because of tumor involvement of both lobes of the liver, anatomy, excessive size of the tumor or secondary extrahepatic metastasis. In unresectable or recurrent cases of the disease, TACE is an important palliative treatment option. TACE is a simultaneous blockage of tumor blood flow by embolizing arteries and localized antitumor effects by delivering chemotherapy directly into the tumor, resulting in dual effects: Cutting off the supply of nutrients and killing cancer cells directly[13]. Nevertheless, the very interventional procedure can worsen liver dysfunction and cause numerous complications related to treatment. The postoperative complication in 386 patients with HCC who underwent TACE was examined in this study and the result showed a complication rate of 26.94% (104/386). This was not as low as the 21.62% in Yang et al[14], yet it was much less than the 35.2 in the study of Yuan et al[15] so the complication risk in this group was moderate. Thus, the prevention of postoperative complications after TACE needs to be optimized as an urgent issue in order to improve safety in the treatment.

In this study logistic regression analysis showed that clinical and procedural factors like age 60 years and older, liver cirrhosis, vascular invasion, tumor diameter more than 5 cm, Child-Pugh grade B, TACE procedures more than 2 times, and operative duration more than 2 hours were found to be significant risk factors of postoperative complications which was also consistent with other studies[16-18]. This puts an emphasis on the specific improvement in nursing care: (1) Enhance preoperative functional evaluation and postoperative monitoring of vital signs and systemic functions in elderly patients and patients with liver cirrhosis; (2) Strengthen preoperative imaging assessment awareness, coordination between hemodynamic management in the operating room, and enforced strict bed rest and the use of abdominal belts in patients with vascular invasion or large tumors; (3) Increase awareness of cumulative injury risks in patients with case management, schedule optimization of intervention and rest between interventions; and (4) Intensive management of drug metabolism and hydration therapy in Child-Pugh B-grade patients.

Nutritional status [preoperative NRS 2002 Score ≥ 3, odds ratio (OR) = 1.362] is an independent risk factor of postoperative complications among the indicators related to nursing. Malnutrition may lead to weakening of immunity, delay in the healing of tissues, and risk of infections[19]. It is suggested that NRS should be a part of regular nursing examination. In patients at high risk, tailored nutritional care (e.g., adjustment of the diet, enteral/parenteral nutrition) must be implemented early, and the nutritional parameters must be dynamically monitored to achieve proper preoperative nutritional status, which helps to protect the postoperative recovery. The risk of complications is significantly increased by self-care capacity and quality of life (low postoperative ESCA Score, OR = 1.677; low QLICP-LI Score, OR = 1.565). Self-management efficacy and health perception of patients are critical towards postoperative recovery. Systematic health education (e.g., postoperative self-management, medication guidance) should be practiced in nursing to promote patient engagement and compliance. Psychological status (preoperative SAS ≥ 50, OR = 1.887; SDS ≥ 53, OR = 1.745) are also a significant risk factor. Negative emotions can affect adherence and inhibit immunity, slow down recovery.

Preoperative psychological screening of the nursing practice should be performed regularly. In patients with anxiety or depression, liaise with psychiatric units to carry out personalized approaches to such problems, including cognitive behavioral therapy and relaxation training, which may be supported by prescription of medications as needed to help create a favorable psychological climate. The impact family support has on the quality of recovery is immense as caregiver burden (high postoperative ZBI Score, OR = 2.119) is a significant predictor of complications. High burden of caregivers undermines patient compliance and quality of nursing care. Educational guidance, social resource connection, and emotional support should be offered to caregivers by nursing to evaluate the stress of caregiving early postoperatively in order to maximize the capacity of caregiving and improve the family support system. Also, the lack of adherence to early postoperative activities and functional exercises (compliance < 80, OR = 1.640; exercise completion < 80, OR = 2.748) is a significant risk factor that puts people at risk of complications. Any prolonged bed rest and activity restriction may precipitate thrombosis, infection and other complications[20].

These results imply that nursing evaluations cannot be limited to the conventional clinical indicators. They are, instead, advised to combine multidimensional aspects including psychological health, nutritional health, family support and rehabilitation behaviors to develop a systemized and personalized risk evaluation system. In patients with anxiety or depression, psychological intervention needs to be incorporated in the early stages of their care and treatment to improve the adherence to the treatment and the immune functioning. In families with high care giving loads, nursing professionals are advised to take the initiative to provide skills training, emotional support, and links to social support to reduce care stress, thus indirectly enhancing patient recovery outcomes.

The risk prediction model developed in the current study based on the independent risk factors mentioned above showed a predictive performance AUC of 0.936 of post-interventional complications through receiver operating characteristic curve analysis with sensitivity and specificity of 93.25% and 84.96, respectively. This shows that the model has a strong discriminative power and high validity that will give nursing personnel a valid instrument to use to stratify risks of complication. This model has many benefits compared to the traditional single clinical indicators or empirical judgments. One, it had a higher C-index of 0.773 (95% confidence interval: 0.746-0.853) and AUC of 0.936 and thus a better predictor of high-risk patients than most of its equivalents. To the patients who are defined high-risk according to the model, the nursing teams can take the initiative of improving the preoperative preparations, maximizing the intraoperative monitoring guidelines, improving the postoperative monitoring rates, and developing specific contingency measures. Also, the model addresses the constraint of the past studies that only dealt with non-dynamic parameters of the procedures, including operational time and TACE operations. This renders risk assessment closer to the demands of clinical nursing practice which are dynamic. But the cases in this study were all obtained at one center, which is a source of possible selection bias. Past research has largely concentrated on clinical risk factors of postoperative complications after TACE (such as liver function status and tumor characteristics), or have studied their effects on prognosis only based on individual nursing aspects, such as nutrition or psychology.

Three main features of the study contribute to its innovation: (1) Integration: This is the first study that includes clinical (vascular invasion, Child-Pugh classification) and treatment (procedure duration, TACE sessions) and multiple nursing (psychological status, nutritional status, self-care ability, family burden, and rehabilitation behavior) factors into a predictive model, which systematically reflects the multifactorial interactive processes that occur in the development of complications; (2) Practicality: The model has a high predictive efficiency (AUC = 0.936) and a good calibration. All the indicators included are easily accessible in clinical nursing practice, which guarantees high translational applicability; and (3) Early warning value: The model emphasizes the independent predictive value of nursing factors that can be controlled (e.g., early compliance with activities, the rate of functional exercise completion), which can give a clear idea of the time and focus of nursing interventions. This is in response to the shortcomings of the past models which focused too much on biological indicators which are not modifiable.

CONCLUSION

Overall, the risk prediction model created in the given study is a useful tool that allows the nursing staff to determine the risk of complications after interventional therapy of liver cancer and introduce the concept of tiered nursing care. It is suggested that this model should be combined with dynamic monitoring of the liver work of patients in the clinical nursing practice and be included in the preoperative nursing examination system. By combining liver function monitoring with nursing assessment scales, a comprehensive “prediction-early warning-intervention” nursing management pathway can be established. This approach facilitates the early identification of high-risk patients, optimizes the allocation of nursing resources, enhances nursing quality, and improves patients’ quality of life.