Early intensive antihypertensive treatment in high-risk population of intracerebral haemorrhage expansion identified by artificial intelligence (ARCHES): study protocol for a multicentre randomised controlled trial

Abstract

Backgrounds

Haematoma expansion (HE) is a significant factor in poor outcomes following intracerebral haemorrhage (ICH). Studies have suggested that acute intensive antihypertensive treatment could reduce HE. However, the impact of early intensive blood pressure reduction on patients with ICH at high risk of HE remains unclear. Therefore, screening ICH patients for high risk of HE upon admission and initiating early intensive blood pressure reduction could improve their prognosis. In this study we utilise a five-point scoring system integrating a deep learning system based on non-contrast CT imaging and clinical predictors to identify ICH patients at high risk of HE. This study aims to compare the efficacy, safety, and feasibility of early intensive antihypertensive treatment versus standard antihypertensive treatment for patients identified as being at high risk of HE.

Methods

The early intensive antihypertensive treatment in high-risk population of intracerebral haemorrhage expansion predicted by artificial intelligence (ARCHES), is a multicentre, prospective, randomised, open-label, blinded-endpoints clinical trial that will include an estimated 680 participants. ICH patients within 6 h of symptom onset, and at high risk of haematoma expansion with ≥ 3 points on the artificial intelligence-based haematoma expansion 5-point prediction score, will be randomly assigned to receive either intensive antihypertensive treatment (targeting systolic blood pressure control between 130 and 140 mmHg within the first hour of treatment, and maintaining this level for 7 days) or standard antihypertensive treatment (targeting systolic blood pressure control between 140 and 180 mmHg, and maintaining for 7 days). The primary outcome is death or severe disability at 90 days.

Discussion

The ARCHES study aims to verify the hypothesis that early intensive blood pressure lowering leads to reduced HE and improved functional outcomes with good safety in ICH patients at high risk of HE.

Trial registration

This study was registered at the Clinical Trials under registry number NCT06242938. Registered on Feb 2, 2024. https://clinicaltrials.gov/study/NCT06242938.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13063-026-09833-x.

Introduction

Background and rationale

Acute hypertensive reaction after intracerebral haemorrhage (ICH) is significantly associated with increased risk of adverse outcomes, including haematoma expansion (HE), early neurological deterioration (END), severe disability, and death [1, 2]. Consequently, the strategy for blood pressure management in the acute stage of ICH is imperative.

A meta-analysis based on the data of Intensive Blood-Pressure Lowering in Patients with Acute Cerebral Haemorrhage II (ATACH-II) [3] and Rapid Blood-Pressure Lowering in Patients with Acute Intracerebral Hemorrhage-2 (INTERACT-2) [4] had preliminarily confirmed the safety of early intensive antihypertensive therapy, but its efficacy still had remained controversial. Another meta-analysis [5] of 5 clinical studies showed that intensive BP lowering in acute ICH was generally safe but did not reduce 90-day mortality or disability. Yet another meta-analysis [6] involving 16 clinical trials demonstrated that intensive blood pressure control significantly reduced HE at 24 h after onset but did not improve modified Rankin Scale (mRS) at 90 days compared with placebo or guideline treatment, and the effectiveness of intensive blood pressure control varied according to blood pressure control strategy and medication selection. Recent studies of intensive blood pressure reduction in ICH patients have updated our knowledge. The INTERACT 3 trial [7] showed an intensive care bundle with blood pressure reduction resulted in improved functional outcomes for patients with acute ICH. However, it remains unclear whether intensive blood pressure reduction alone can improve outcomes. The INTERACT 4 trial [8] suggested that prehospital blood pressure reduction improved functional outcomes of ICH patients, whereas it did not in ischaemic stroke patients. In addition, the elaborate care bundle protocol and prehospital CT scan are difficult to implement on a large scale, and become limitations of the above studies. ICH patients who develop HE are more likely to face a catastrophic outcome. Early intensive antihypertensive therapy has the potential to improve their prognosis, but there is a lack of evidence of effectiveness. Thus, whether patients at high risk of HE could benefit from early intensive antihypertensive therapy warrants further investigations.

Objectives

In the proposed ARCHES trial, we aim to validate the hypothesis that early intensive antihypertensive treatment with an SBP target of 130–140 mmHg in ICH patients at high risk of HE will reduce HE (baseline to follow-up CT scan at 24 h > 6 ml) and decrease the occurrence of death or severe disability (mRS score ≥ 3) at 90 days, compared with those receiving the standard antihypertensive treatment with an SBP target of 140–180 mmHg.

Trial design

The proposed ARCHES trial is a multicentre, prospective, randomised, open-label, blinded-endpoints trial (PROBE). Randomisation will be 1:1 for intensive antihypertensive treatment versus standard antihypertensive treatment via a centralised website. Patients will be followed up for 180 days. BP measurements and other relevant data will be collected at baseline, 24 h of onset or END, on 7–10 days, at discharge during hospitalisation, and on 90 days and 180 days upon onset during the follow-up visit. Approximately 78 centres in China will participate. Estimated study duration is 11 months. This study was registered at the Clinical Trials under registry number NCT06242938. The protocol of this study was developed according to the Standard Protocol Items Recommendations for Interventional Trials (SPIRIT) guidelines, with the SPIRIT checklist included as an Additional file. A detailed diagram of the methodology used in this study is shown in Fig. 1.

Fig. 1.

Study methodology. Abbreviations: END, early neurological deterioration; NIHSS, National Institute of Health stroke scale; mRS, modified Rankin scale; EQ-5D, EuroQol Five Dimensions Questionnaire; PHQ-9, Patient Health Questionnaire-9

Methods

Participants, interventions and outcomes

Study setting

The study will be conducted at 78 local hospitals in China (Additional file). Each centre has a trained principal investigator responsible for recruitment. Regular investigator meetings and monthly enrolment reports will be used to monitor recruitment progress.

Research implementation period

The study was planned to be conducted from February 15, 2024 until the scheduled end of January 31, 2025.

Eligibility criteria

Inclusion criteria:

1. Age ≥ 18.

2. Arrival at the hospital within 6 h of onset and be able to complete randomisation and receive antihypertensive therapy within 1 h of arrival (if the time of onset is unclear, the latest normal time is applicable).

3. CT-confirmed spontaneous supratentorial parenchymal haematoma with (or without) extending into the ventricles or subarachnoid space.

4. Haematoma volume < 60 ml.

5. Artificial intelligence-based HE 5-point prediction score (termed as ARCHES) for patients at high risk of HE, with a score ≥ 3. The ARCHES score was defined as Ai-Reinforced intraCerebral Haemorrhage hematoma Expansion Score, a 4-item prediction algorithm with a total score ranging from 0 to 5. Individual components of the ARCHES score are shown in Table 1.

6. Glasgow coma scale (GCS) > 8.

7. Any measurement of SBP ranging from 150 to 220 mmHg from onset to randomisation.

8. Signed informed consent by the patient or legal representative.

Table 1.

Individual components of the ARCHES score

Variables	Cut-off	Points
Antihypertensive therapy history	No	1
	Yes	0
Onset-to-CT time	≤ 3.5 h	1
	> 3.5 h	0
Baseline NIHSS	≥ 12	1
	< 12	0
DLS prediction probability	< 0.36	0
	0.36–0.64	1
	≥ 0.64	2

DLS, deep learning HE prediction system based on non-contrast CT; NIHSS, National Institute of Health stroke scale

Exclusion criteria:

1. Cerebral haemorrhage caused by known secondary factors such as trauma, tumours, cerebral vascular malformations, thrombolysis and mechanical thrombectomy.

2. Haemorrhage of the brain parenchyma involving the ventricles when blood completely fills one lateral ventricle or more than half of both lateral ventricles.

3. Previous intracerebral haemorrhage with unabsorbed haematoma.

4. Surgical intervention planned within 24 h.

5. Combination of conditions known to contraindicate intensive blood pressure lowering (e.g. severe carotid artery stenosis, vertebral or intracranial artery stenosis, Moyamoya disease, Takayasu’s arteritis, and severe cardiac valve stenosis)

6. Combination of known well-defined indications requiring a more aggressive antihypertensive regimen, such as hypertensive encephalopathy or aortic dissection.

7. Known allergy to antihypertensive medications

8. Patients who have received warfarin with an INR > 1.5 or other anticoagulants in the past 24 h.

9. Subjects with platelet counts less than 50,000/mm³.

10. Disability due to prior illness with an mRS ≥ 3.

11. Pregnancy status or within 30 days of delivery.

12. Severe cardiac or hepatic disease, severe renal insufficiency (eGFR < 30 ml/min) or malignancy with life expectancy < 3 months.

13. Currently participating in other interventional clinical trials.

14. Inability to obtain informed consent.

Ethics and informed consent

This trial will adhere to the ethical standards outlined in the Declaration of Helsinki and the guidelines for medical and health research involving human participants. The ethical review boards of all involved hospitals approved the study protocol (main ethics Institutional Review Board approval: KY2023-138-02). Written informed consent was obtained from all participating patients and their legal proxies.

Randomisation and allocation concealment

Eligible patients will be randomised to receive either intensive antihypertensive treatment or standard antihypertensive treatment (1:1 at each study site). Block randomisation was done with the use of a central web-based randomisation system (Randomisation and Trial Supply Management version 3.1.2, Beijing Bioknow Information Technology, China) with a block length of two or four without stratification. The local investigators visited the web-randomisation system and obtained the random codes, and the treatment assignment was done according to the random code.

Trial interventions

This study will compare intensive antihypertensive and standard antihypertensive therapy for ICH patients at high risk of HE. The antihypertensive treatment is open-label, while investigators involved in the subsequent clinical and imaging assessment of outcomes will remain blinded to treatment allocation.

In addition to antihypertensive therapy, both groups will receive the same standard management in accordance with current guidelines.

Intensive antihypertension group

Antihypertensive treatment will be given within 1 h after randomisation. SBP will be controlled to 130–140 mmHg within 1 h after antihypertensive treatment and maintained at this level for 7 days. The physicians select antihypertensive medications on their own, but should follow the principle of intravenous combined oral antihypertensive and select antihypertensive medication under the guidance of this research.

Guidance of antihypertension medication:

1. Intravenous antihypertensive medications:

   • Intravenous antihypertensive medications (e.g. uradil and nicardipine) need to be combined with sufficient oral antihypertensive drugs concurrently.
   • It is recommended that the intravenous antihypertensive drugs be reduced and stopped within 3–7 days.
2. Blood Pressure Reduction:

   • For SBP 140–179 mmHg at randomisation, required to reach the target range within 1 h.
   • For SBP 180–220 mmHg at randomisation, reduced to 150–160 mmHg within 1 h, required to reach the target range within 6 h.
3. Oral antihypertensive medication regimen:

   • For SBP 150–159 mmHg: double oral antihypertensive medications are recommended.
   • For SBP 160–179 mmHg: triple oral antihypertensive medications are recommended.
   • For SBP 180–220 mmHg: quadruple oral antihypertensive medications are recommended.
   • If the patient was already on antihypertensive medication with an SBP of 140–149 mmHg prior to randomisation and intervention medication, dual oral antihypertensive medications are still recommended.
   • The addition of a beta-blocker is recommended when the heart rate is > 80 beats/minute.
4. Available antihypertensive medication classes (specific dosing instructions are provided in the form of pamphlet or applications):

   • α-blockers (e.g. terazosin and doxazosin),
   • β-blockers (e.g. labetalol, metoprolol, and bisoprolol),
   • α/β-blockers (e.g. arnolol and carvedilol),
   • Calcium channel blockers (CCBs) (e.g. nifedipine, behenidipine, amlodipine, etc.),
   • Diuretics (e.g. hydrochlorothiazide, spironolactone, and furosemide),
   • Angiotensin receptor blockers (ARBs)/angiotensin-converting enzyme inhibitors (ACEIs) (e.g. irbesartan, valsartan, temesartan, cloxartan, enalapril, benazepril, fosinopril, and perindopril). ANRI (sacubitril valsartan)

Standard antihypertension group

SBP will be controlled to 140–180 mmHg after randomisation and maintained for 7 days. The physicians are allowed to choose antihypertensive medications based on their own clinical experience.

Assessment and follow-up

The schedule of assessments is shown in Fig. 2. Clinical assessments will be performed during hospitalisation and after discharge. In this trial, direct follow-up by the principal investigator and sub-investigators is the primary approach. For participants who miss follow-up visits, efforts will be made to collect data via telephone contact whenever possible. For participants who discontinue intervention, change their intervention regimen, or withdraw from the study early, we will continue to collect follow-up data to the extent possible, provided informed consent has been obtained. In accordance with the intention-to-treat (ITT) principle, all participants who have been randomised to the study will be included in the primary analysis according to their original assigned group. Additionally, participants who did not deviate from the protocol will be included in the per-protocol set (PPS) for sensitivity analysis.

Fig. 2.

Schedule of assessments. *1: Record blood pressure and heart rate: < 1 h: every 15 min (for the first 15 min, every 5 min);1–6 h: every 30 min;6–24 h: every 4 h;24 h–7 days: every 8 h. *2: Laboratory tests, including complete blood count, CRP, liver function (ALT, AST, albumin), renal function panel (blood urea nitrogen, creatinine, GFR, glucose), and electrolytes (3 items). Laboratory tests on 7–10 days, including complete blood count, CRP, coagulation profile (including D-dimer), liver function (ALT, AST, albumin), renal function panel (blood urea nitrogen, creatinine, GFR, glucose), electrolytes, routine urine test and glycated haemoglobin (can be completed during hospitalisation). *3: In the event of END, a repeat head CT scan is required. The CT scan should be performed with a slice thickness of 5 mm, providing continuous coverage of the entire brain, with 28 to 32 slices captured. *4: Serious adverse events (SAE), refer to the Common Terminology Criteria for Adverse Events (CTCAE) version 5.0, levels 3 and above

BP measurements

BP will be measured before randomisation as a baseline record. After the antihypertensive treatment is given, BP measurements will be conducted at the following intervals: every 5 min for the first 15 min, every 15 min from 15 min to 1 h, every 30 min from 1 to 6 h, every 4 h from 6 to 24 h, and every 8 h from 1 to 7 days. Each BP assessment will require two brachial BP measurements, 30 s apart, utilising an automated measurement device (Omron HBP-1300 Professional BP Monitor) with patients in the supine position. The average measures will be recorded.

Outcome measures

Primary outcome

The primary outcome is a composite of death and major disability (mRS ≥ 3) at 90 days.

Secondary outcomes

Efficacy outcomes:

1. HE within 24 h (HE is defined as a volume increase > 6 mL from baseline to follow-up non-contrast CT)

2. END within 24 h defined as a ≥ 2 reduction in GCS or a ≥ 4 increase in NIHSS score for non-sedatives/sleeping medications within 24 h of onset (compared to baseline).

3. Ordinal distribution of mRS at 90 days.

4. Death within 90 days.

5. Quality-of-life EuroQol-5 Dimensions (EQ-5D) at 90 days.

6. Proportion of death and major disability at 180 days.

Safety outcomes:

1. Severe hypoperfusion event (e.g. acute kidney impairment, severe hypotension, acute myocardial infarction, and acute cerebral infarction) during hospitalisation.

2. Treatment-related severe adverse events (SAE) during hospitalisation.

3. Death from any cause during hospitalisation.

Sample size

There were two arms, the intensive antihypertension group and the standard antihypertension group. The group allocation ratio was 1:1. Based on data from the previous INTERACT 2 study [4] and relevant AI prediction models, the assumed event rate of clinical poor outcomes (mRS 3–6 points) at 90 days in the standard antihypertensive treatment group within the AI-predicted high-risk population for HE is 55.6%. It is anticipated that the intensive antihypertensive treatment could reduce the event rate by 11%. The total sample size of 680 for the study will yield 80% power to observe an 11% difference in proportions of patients mRS 3–6 points (44.6% intervention vs 55.6% control), assuming two-sided alpha of 0.05, an attrition rate of 5%.

Statistical analyses

Data will be analysed according to intention-to-treat principle. Differences between study groups in the rate of 90-day mRS 3~6 will be assessed with the use of a generalised linear model with the trial centre as a random effect. The relative rate (RR) and the 95% CI will be reported. All effects of Intensive antihypertension group, as compared with standard antihypertension group, on binary secondary and safety outcomes will be reported as RRs with respective 95% CIs. The distribution of scores on the modified Rankin scale will be analysed with the use of ordinal logistic regression and will be reported as a common odds ratio (OR); verification of the proportional-odds assumption will be performed with the use of a chi-square test. Time-to-event analysis will be employed to compare the cumulative hazards of death within 90 days between the intensive antihypertensive treatment group and the standard antihypertensive treatment group. Kaplan–Meier curves on cumulative hazards by follow-up time will be examined and the group differences will be tested by a log-rank test. Cox regression will be utilised to obtain hazards ratios (HRs) and 95% CIs. For the analysis of EQ-5D-5L scores at 90 days, t test or Wilcoxon rank-sum test will be used and we will calculate the differences in scores between groups and their 95% CIs.

For comparison between groups, all hypothesis tests will be performed with a two-sided alpha level of 0.05 (α = 0.05). Multiple imputation will be used to address the issue of missing data. Furthermore, as a sensitivity analysis for the primary outcome of 90-day mRS, missing values will be imputed as 3 to 6. All statistical analyses are processed by SAS V.9.4 or higher version statistical software and will be performed on a predetermined statistical analysis plan.

Trial oversight

This study will be conducted at Beijing Tiantan Hospital and 77 participating sub-centres. The data centre is co-located with data management personnel to facilitate overseeing centralised data monitoring throughout the trial period. The Operations Committee played a role in formulating the protocol and will supervise the progression of the trial (Table 2). No patients or members of the public were involved in the design of this study. The principal investigator reviews cumulative enrolment each month. An investigator kick-off meeting was held at study launch, and the principal investigator remains in close contact with participating sub-centres by email or teleconference so that even minor operational issues can be resolved promptly.

Table 2.

Steering committee

Role in this study	Name	Institution
Principal investigator	Na Li	Department of Neurology, Beijing Tiantan Hospital
Supervision/overall guidance	Xingquan Zhao	Department of Neurology, Beijing Tiantan Hospital
Research support	Jiandong Yu	China National Clinical Research Center for Neurological Diseases
Research collaborator	Wanxing Ye	China National Clinical Research Center for Neurological Diseases
Event adjudication committee	Qi Li	Department of Neurology, The Second Affiliated Hospital of Anhui Medical University
Event adjudication committee	Hui Qu	Department of Neurology, Beijing Tiantan Hospital
Event adjudication committee	Xia Meng	Department of Neurology, Beijing Tiantan Hospital
Study statistician	Anxin Wang	Department of Statistic, China National Clinical Research Center for Neurological Diseases
Study secretariat		Beijing Tiantan Hospital
Project management		Beijing Tiantan Hospital

In case of any protocol amendments, the principal investigator will notify all participating sub-centres of the changes. A copy of the revised protocol will be filed in the Investigator Site File at each site. Any protocol deviations will be documented using a breach report form. The clinical trial registry entry will be updated accordingly.

Monitoring and audit

In order to ensure the thoroughness and precision of data collection and entry, a rigorous quality control process has been implemented. The completion and verification of case report forms is the responsibility of the trained clinical research coordinators (CRCs). The data entry phase will involve dual independent data entry with cross-validation for data derived from paper-based case report forms. Automated logic checks are preset for the electronic data capture (EDC) system to identify outliers or logical inconsistencies. In this trial, monitoring will be conducted by the data and safety monitoring board (DSMB) of China National Clinical Research Center for Neurological Diseases. Biannual routine monitoring sessions will be performed based on the monitoring reports prepared by the Research Administration Office. These sessions involve a comprehensive meeting with all or some representatives of group leaders, principal researchers, and administrative offices, the Efficacy and Safety Evaluation Committee, and participating institutions. Routine on-site monitoring will occur at least once every 3 months to review study progress and protocol compliance, with additional visits triggered by SAEs, major protocol deviations, or other emerging issues. Key source data verification will include informed consent, eligibility, primary and secondary endpoints, SAEs, and protocol compliance. Adherence will be verified during these visits and during independent audits. All anticipated adverse events will be coded according to the Medical Dictionary for Regulatory Activities (MedDRA) and graded according to Common Terminology Criteria for Adverse Events (CTCAE) to ensure consistent classification and analysis throughout the study.

Auditors will visit the representative facilities of the trial to verify the facility’s Institutional Review Board (IRB) documents and patient consent forms and reconcile the data on electronic case report forms (CRF) on the data capture (EDC) system and source medical records. Auditing is scheduled to take place every 6 months. The audit is independent of the sponsor and investigators.

Ancillary and post-trial care

The study will cover the cost of treatment for any injury or complication that is a direct consequence of participation in this clinical trial. Standard of care for the subject’s underlying medical condition, and treatment for any medical conditions unrelated to the trial, are the responsibility of the subject/subject’s national health insurance system.

Confidentiality

A designated party is responsible for storing all study-related information and materials; this responsibility includes the retention of records relevant to the study (copies of application documents, notification documents from the director, various application forms and reports, study-related materials, and other documents or records necessary to ensure data reliability). Furthermore, personal information will be protected during the disposal of documents or materials by shredding and incinerating physical documents (for electronic data, this involves using dedicated data deletion software or physical destruction to render the data irrecoverable). The principal investigator will take measures to ensure that these documents are not accidentally or prematurely destroyed. Retention period: Documents will be retained for 10 years after the study’s end is reported or for 8 years after the final results are published, whichever period is longer.

Access to final trial dataset

The final trial dataset, including the anonymised individual participant data, will be under the control of the Sponsor. The principal investigator for the overall trial and the study statisticians will have direct access to the dataset to perform the final analysis. There are no contractual agreements that limit access to the final dataset for the principal investigator. The principal investigator will be granted access to the relevant aggregated data and the clinical study report to facilitate the review and interpretation of the results for publication and regulatory purposes.

Discussion

The 2022 American Heart Association (AHA) Guidelines for the Management of Spontaneous Cerebral Haemorrhage [9] assert that in patients with mild-to-moderate ICH and an initial systolic blood pressure of 150–220 mmHg, lowering the systolic blood pressure to 130–140 mmHg at the acute stage is safe and may be reasonable in improving functional outcome. However, the optimal BP control strategy remains to be discussed in depth.

A pooled analysis [10] of INTERACT2 and ATACH-II data showed in 3796 ICH patients within 4.5 h of onset of symptoms that a J-shape relationship between SBP levels within 1 h after randomisation and poor functional outcomes at 90 days of onset. However, excessive BP reduction may be harmful to ICH patients, especially by increasing cardio-renal adverse events caused by hypoperfusion [11]. The ARCHES trial excludes patients with an initial SBP of > 220 mmHg, and patients will accept a step-down BP lowering treatment according to different level of initial SBP (SBP 150–179 mmHg and 180–220 mmHg) in the intensive antihypertensive treatment group. This approach may avoid excessive BP reduction over a short period. The earlier the antihypertensive treatment starts, the more likely the ICH patient is to benefit. Ultra-early (within 2 h after onset) intensive antihypertensive treatment significantly reduced the risk of HE and improved the 90-day functional outcomes in ICH patients [12]. The latest INTERACT 4 trial has demonstrated that reducing systolic blood pressure to 130–140 mmHg within 2 h of onset improves 90-day functional outcomes in patients with ICH but leads to catastrophic consequences in patients with cerebral infarction [8]. Consequently, there is a necessity to identify patients with ICH by completing a CT scan in a prehospital ambulance. However, this process may prove challenging to achieve. The INTERACT 3 trial showed that the implementation of a goal-directed care bundle, which included early intensive blood pressure reduction, strict glucose control, antipyretic treatment and rapid reversal of anticoagulation in hospital protocols, resulted in improved outcomes for ICH patients presenting within 6 h of the onset of symptoms [7]. However, implementation of this goal-directed care bundle into routine practice may also be challenging [13].

Previous reported clinical trials targeting HE mostly applied an ‘all comers’ approach [3, 4], without selection of patients in high-risk HE based on clinical and radiological characteristics. ‘All comers’ approach might have diluted the real beneficial effect of targeted treatment in a large pool of patients, for the specific populations at high risk of HE more likely to benefit from intensive antihypertensive therapy. Our ARCHES trial will target ICH patients at high risk of HE, screen them on arrival and administer intensive antihypertensive therapy to confirm its safety and efficacy in a more accessible way.

The optimal selection tool for identifying patients at risk of HE should be easy to evaluate and accurate. Imaging markers for HE risk prediction, such as CTA ‘spot sign’, NCCT signs, and BAT score, have been widely studied [14–16]. Although CTA ‘spot sign’ and NCCT signs are able to predict HE, there is no evidence that intensive antihypertensive therapy benefits patients at high risk of HE suggested by these markers due to their generally poor sensitivity [17, 18]. We developed a five-point prediction ARCHES score combining a deep learning system (DLS) HE prediction model based on NCCT imaging and clinical variables. Internal and external validations have demonstrated that the ARCHES score exhibits better sensitivity and specificity in identifying individuals at high risk of HE than previous HE prediction models. In the test dataset, the AUC of the ARCHES was 0.760 (95%CI 0.662–0.841), still significantly superior to both the BAT score [15] (AUC, 0.639; 95%CI 0.535–0.735; P = 0.042) and the meta-analysis prediction model [19] (AUC, 0.603; 95%CI 0.498–0.702; P = 0.048). In the ARCHES trial, the easy-to-use five-point ARCHES score will be employed to identify patients at high risk of HE. The application of this accurate and rapid HE prediction model for patient selection may be a chance and breakthrough for intensive antihypertensive therapy in patients with high-risk HE after acute ICH.

Trial status

The protocol was initially on version 1.0 when patient recruitment began on February 15, 2024, and the first patient was recruited on April 10, 2024. During the recruitment phase, the protocol was not updated. The trial is scheduled to end on January 31, 2025. Due to the unfortunate and repeated rejection of manuscripts previously submitted to other journals, the submission to this journal was consequently delayed. The first follow-up appointment for the last patient in our study is scheduled for January 5, 2026.

Dissemination policy

The trial results will be disseminated within 24 months of completion through peer-reviewed publication and clinical trial registry entry, with a lay summary available to participants. Authorship will adhere to ICMJE criteria, and no professional writers will be used. The datasets used and/or analyzed in the current study are available from the corresponding author upon reasonable request.

Authors’ contributions

NL researched literature and led the proposal and protocol development. NL, LW, SS, KK, XM, HL and QX were involved in protocol development, gaining ethical approval, patient recruitment and data collection. AW and WY did the statistical analysis. LW and SS wrote the first draft of the manuscript. WY, XZ and NL revised this manuscript. NL and XZ did the funding acquisition, resources and supervision. All authors reviewed and edited the manuscript and approved the final version of the manuscript.

Funding

This work was supported by National Key R&D Program of China (2022YFC2504903, 2017YFC1310901, 2016YFC0901002, 2017 YFC1307905 and 2022ZD0118005); Beijing Municipal Committee of Science and Technology (Z201100005620010); National Natural Science Foundation of China (81371293); Beijing Nova Program (Z131107000413042). The sponsor played no part in study design; collection, management, analysis, and interpretation of data; writing of the report; and the decision to submit the report for publication.

Data availability

Not applicable, as this manuscript is a study protocol.

Declarations

Ethics approval and consent to participate

The protocol and consent forms were approved by the ethics committees of Beijing Tiantan Hospital and each participating institution. Written informed consent was obtained from all participating patients and their legal proxies.

Competing interests

The principal investigator for the overall trial affirms that they have no financial or other competing interests that could inappropriately influence, or be perceived to influence, the design, conduct, analysis, or reporting of this trial. For each study site: The principal investigators at all participating study sites have similarly declared that they have no financial or other competing interests related to this trial.