Machine Learning Techniques in Blood Pressure Management During the Acute Phase of Ischemic Stroke

 

So we still have NO FUCKING CLUE what a blood pressure management protocol is. Hope you don't mind dying because of the cesspools of incompetence of the complete stroke medical world.  Unless YOU hold your stroke hospital's feet to the fire you are allowing your children and grandchildren to die or become disabled.

And with no measurement of 100% recovery they are not even trying to solve stroke.

Machine Learning Techniques in Blood Pressure Management During the Acute Phase of Ischemic Stroke

Orit Mazza^1,2*, Onn Shehory¹ and Nirit Lev^3,4,5

• ¹Graduate School of Business Administration, Bar Ilan University, Ramat Gan, Israel
• ²Lowenstein Rehabilitation Medical Center, Ra'anana, Israel
• ³Neurology Department, Meir Medical Center, Kfar Saba, Israel
• ⁴Sackler Faculty of Medicine, Tel-Aviv University, Tel-Aviv, Israel
• ⁵Sagol School of Neuroscience, Tel-Aviv University, Tel-Aviv, Israel

Background and Purpose: Elevated blood pressure (BP) in acute ischemic stroke is common. A raised BP is related to mortality and disability, yet excessive BP lowering can be detrimental. The optimal BP management in acute ischemic stroke remains insufficient and relies on expert consensus statements. Permissive hypertension is recommended during the first 24-h after stroke onset, yet there is ongoing uncertainty regarding the most appropriate blood BP management in the acute phase of ischemic stroke. This study aims to develop a decision support tool for improving the management of extremely high BP during the first 24 h after acute ischemic stroke by using machine learning (ML) tools.

Methods: This diagnostic accuracy study used retrospective data from MIMIC-III and eICU databases. Decision trees were constructed by a hierarchical binary recursive partitioning algorithm to predict the BP-lowering of 10–30% off the maximal value when antihypertensive treatment was given in patients with an extremely high BP (above 220/110 or 180/105 mmHg for patients receiving thrombolysis), according to the American Heart Association/American Stroke Association (AHA/ASA), the European Society of Cardiology, and the European Society of Hypertension (ESC/ESH) guidelines. Regression trees were used to predict the time-weighted average BP. Implementation of synthetic minority oversampling technique was used to balance the dataset according to different antihypertensive treatments. The model performance of the decision tree was compared to the performance of neural networks, random forest, and logistic regression models.

Results: In total, 7,265 acute ischemic stroke patients were identified. Diastolic BP (DBP) is the main variable for predicting BP reduction in the first 24 h after a stroke. For patients receiving thrombolysis with DBP <120 mmHg, Labetalol and Amlodipine are effective treatments. Above DBP of 120 mmHg, Amlodipine, Lisinopril, and Nicardipine are the most effective treatments. However, successful treatment depends on avoiding hyponatremia and on kidney functions.

Conclusion: This is the first study to address BP management in the acute phase of ischemic stroke using ML techniques. The results indicate that the treatment choice should be adjusted to different clinical and BP parameters, thus, providing a better decision-making approach.

Introduction

Machine learning (ML) applications in healthcare have significant potential for improving clinical decision-making diagnoses, treatment effectiveness, and healthcare management, including lowering the costs for both healthcare providers and patients (1). ML applications for Knowledge Discovery in Databases (KDD) have been used for more than two decades and are useful for discovering information and extracting knowledge from data and reflect a multi-step process that involves thorough data preparation, pattern searching, and knowledge evaluation (2). The use of ML to extract non-trivial and previously unknown useful information from data may be most beneficial for physicians in areas where the level of evidence or class of recommendation is low and will increase the likelihood of the physicians adopting them (3). The use of ML in clinical research to predict a particular clinical outcome is useful because it has the potential to outperform the best clinical knowledge obtained by current traditional medical research. In this study, we applied the KDD process by using ML techniques to conduct a robust interrogation to identify predictors of blood pressure (BP) management after acute ischemic stroke, thus, having the potential to aid clinicians in improving treatment regimens.

An elevation in BP is common in the acute phase of a stroke and occurs early at the time of arrival to the emergency room. In two-thirds of the patients, elevated BP was transient and resolved within 2 weeks from symptom onset (4). Observational studies have shown that elevated BP during ischemic stroke onset is prognostically associated with an increased risk of early adverse events and mortality. However, acute and aggressive BP lowering within 24 h of stroke onset could also jeopardize the outcome (5). Both elevated and low BP are independent factors that predict poor outcomes among patients with acute ischemic stroke and present a U-shaped relationship between BP and death or disability (6, 7).

High BP in acute stroke can decrease blood perfusion to areas of ischemic brain tissue, which, in turn, can cause neurological damage (8). An extremely high BP can result in intracerebral bleeding and hypertensive emergencies, including renal failure, ischemic heart disease, and pulmonary edema (9). In patients who received thrombolytic treatment, studies concur that there is a strong association between high BP and worse clinical outcomes, including death, disability, and hemorrhagic transformation (7, 10). The AHA/ASA and ESC/ESH guidelines recommend lowering the BP below 180/105 mmHg in patients receiving thrombolysis in the first 24 h after acute stroke, a strong class of recommendation (class I). In patients not receiving thrombolysis, a clinical judgment is defined as whether to treat hypertension when it exceeds 220/120 mmHg, a weak class of recommendation (class II-b) (11, 12). There is no firm evidence regarding BP management in patients with acute ischemic stroke with a BP lower than 220/120 mmHg, who did not receive thrombolysis (12). The specific interval for BP reduction is not well-established, and the current approach of lowering BP by 15% is considered reasonable by a consensus expert opinion (11, 12). The current recommended approach by the AHA/ASA is to treat with labetalol, nicardipine, or clevidipine when systolic blood pressure (SBP) is over 180–230 mmHg or diastolic BP (DBP) is over 105–120 mmHg. If DBP exceeds 140 mmHg or is not controlled by these treatments, sodium nitroprusside is recommended. However, these recommendations are not based on a strong class of recommendations (11, 13).

There are therapeutic strategies for elevated BP that are not included in the current acute stroke guidelines. In most hypertensive emergencies, intravenous (IV) drug administration is considered, although oral therapy with ACEI/ARBs or beta-blockers is effective in the acute setting of hypertensive emergency because of the activation of the renin system. Besides the medications mentioned above for BP lowering in acute ischemic stroke, other treatment options are utilized in various hypertensive emergencies including metoprolol, esmolol, nitroglycerine, clonidine, and enalaprilat. The duration of action of these treatments ranges from several minutes to several hours and enables dose adjustment according to clinical judgment (12). The use of these medications in the treatment of acute ischemic stroke is required.

Few randomized clinical trials have examined the impact of BP reduction immediately after acute stroke with antihypertensive agents (14, 15). The effects of continuous antihypertensive treatment, in previously known patients who were hypertensive after acute stroke in the Continue Or Stop post-Stroke Antihypertensives Collaborative Study (COSSACS), showed a statistically significant reduction of 13/8 mmHg in BP at 2 weeks in the continuing group compared to the stop group, and no differences emerged between the groups in rates of serious adverse events, 6-month mortality, or major cardiovascular events (14). However, the aforementioned study had inherent limitations due to the complex clinical situation. It was not placebo-controlled, and there was a multiplicity of pre-existing antihypertensive treatments (14). The China Antihypertensive Trial in Acute Ischemic Stroke (CATIS), a randomized clinical trial, compared patients who received antihypertensive treatment to those who discontinued all antihypertensive medications during hospitalization. The treatment aimed to lower SBP by 10–25% within the first 24 h. The primary outcome of death within 14 days after randomization and major disability at 14 days or hospital discharge did not differ between the groups. However, early antihypertensive therapy was associated with a lower rate of 3-month recurrent stroke among patients with a history of hypertension (15).

Several randomized clinical trials have examined the use of specific antihypertensive agents (16, 17). The Controlling Hypertension and Hypotension Immediately Post Stroke (CHHIPS) randomized controlled trial investigated the effect of BP reduction with labetalol and lisinopril vs. placebo in patients with SBP > 160 mmHg. The SBP reduction within the first 24 h was higher in both treatment groups (16). The Intravenous Nimodipine West European Stroke Trial (INWEST) showed a significant decrease in SBP and DBP with nimodipine treatment vs. placebo in the first 48 h (18). Furthermore, several randomized trials have examined the effects of angiotensin receptor blockers (ARBs) on BP reduction in the acute phase of stroke and observed a modest reduction in BP of up to 10/6 mmHg in the treatment group vs. the placebo group (17, 19, 20).

In many clinical trials evaluating BP-lowering, markedly elevated BP ranges (usually > 220/120 mmHg) were excluded (14–16). However, the guidelines concern the treatment of severe hypertension. In addition, no solid data are available to guide the selection of antihypertensive treatment. Accordingly, the main objective of this research was to develop a decision support tool for improving the management of extremely high BP during the first 24 h after acute ischemic stroke by using ML techniques. To date, no published study has used ML techniques to predict BP management in the acute phase of ischemic stroke.

More at link.