AI sheds light on why stroke recovery differs

 WOW!  Getting better at predicting failure to recover! And you think that is a good use of AI. Your comeuppance/screaming when you are the 1 in 4 per WHO that has a stroke  will be soul satisfying.  Delivering 'care', NOT RECOVERY is a fireable offense!

AI sheds light on why stroke recovery differs

22 September 2026 Health and medicine Auckland Bioengineering Institute Researchers are using artificial intelligence to help predict stroke recovery and support more personalised patient care(

NOT RECOVERY!).

The health of a person's brain before a stroke may help explain why people with similar strokes can experience very different recoveries, according to ongoing research at Waipapa Taumata Rau, University of Auckland. By combining artificial intelligence with MRI brain scans, researchers hope to develop tools that help doctors predict recovery earlier and tailor rehabilitation plans for stroke patients. Leading the project is Simi Sasidharan, a PhD student with an extensive background in IT. Using advanced image-processing techniques, the research investigates how brain conditions before and after a stroke influence recovery. Stroke damage alone may not tell the full story, the study finds. Patients with similar-sized lesions in the same brain region can experience very different outcomes. Senior Lecturer and Postgraduate Director of Medical Imaging Dr Sibusiso Mdletshe says the findings are helping researchers better understand the factors that influence recovery after a stroke. "The patients could have a similar lesion in the brain, but the health of that brain is different," he says. So, we're finding that the health of the brain is quite important in determining how the brain will respond after the stroke. Dr Sibusiso Mdletshe Waipapa Taumata Rau, University of Auckland. The project brings together expertise in medical imaging, artificial intelligence and neurology. Working with MRI scans from hundreds of stroke patients, researchers are creating detailed three-dimensional maps of affected brain pathways and training AI models to identify patterns that may predict recovery. Researchers are using the technology to build a clearer picture of how stroke injuries affect different patients. "In simple terms, AI helps us better understand what is happening within the stroke lesion in a patient's brain," he says. "It allows us to map how that lesion is behaving and how it may affect the patient's recovery in the future."

The study analyses data from the Stroke Motor Rehabilitation and Recovery Study at Massachusetts General Hospital in Boston, United States. Researchers collaborated with Dr David Lin, a critical care neurologist and neurorehabilitation specialist, and Julie DiCarlo, programme manager for the hospital's Laboratory for Translational Neurorecovery.

Mdletshe says the team's focus is now on translating the research into a clinically useful tool.

"If the tool we're developing is able to do what we intend, it will help physicians predict how a stroke will behave and support the recovery plan for that patient," he says.

The project forms part of a wider stroke AI imaging research programme led by Associate Professor Alan Wang from the Auckland Bioengineering Institute. Mdletshe is the principal supervisor of the PhD research, with Associate Professor Wang serving as mentor and co-supervisor.

The research is already drawing attention beyond New Zealand. The PhD project is more than halfway through and has resulted in conference presentations in New Zealand and overseas, as well as published research.

The team hopes future studies will include New Zealand imaging data, helping researchers understand whether the patterns they are seeing internationally are reflected here.

Further validation is needed before any AI tool can be used in clinical practice and researchers continue to test the approach with larger datasets.

"The view is that this will impact the care(NOT RECOVERY!)

 of stroke patients in a significant way in the future," Mdletshe says.

The project reflects his wider interest in using medical imaging to improve patient care(NOT RECOVERY!).

He has also co-led, with Associate Professor Peter Jones, the development of the University's new Postgraduate Certificate in Health Sciences specialisation in Point of Care Ultrasound (POCUS), which will welcome its first students in 2027.

Developed in response to growing demand for clinician-performed ultrasound training in New Zealand, the programme will help healthcare professionals build bedside imaging skills that support faster clinical decision-making.

Media contact 
 

Caryn Wilkinson | Media adviser 
M: 027 202 6372 
E: caryn.wilkinson@auckland.ac.nz 

Inflammatory burden index is associated with poor functional outcome and stroke-associated pneumonia in patients with primary brainstem hemorrhage: a retrospective cohort study

 Predicting failure to recover is grounds for firing! PURE INCOMPETENCE DISPLAYED!

Competent researchers would create pneumonia prevention protocols!

You've known of the problem for years and DONE NOTHING; YOU'RE FIRED!

• 11% Stroke-associated pneumonia (18 posts to October 2020)

Stroke research should deliver recovery protocols. You'll be screaming for them when you are the 1 in 4 per WHO that has a stroke!  

Inflammatory burden index is associated with poor functional outcome and stroke-associated pneumonia in patients with primary brainstem hemorrhage: a retrospective cohort study

• 

  Fangjie Luo 1† *

• J

  Jialiang Luo 2†

• X

  Xiaoyun Li 1

• 1. Department of Neurosurgery, The Affiliated Dongguan Songshanhu Central Hospital of Guangdong Medical University, Dongguan, Guangdong, China

• 2. Sun Yat-sen University, Dongguan, Guangdong, China

Abstract

Background: 

Primary brainstem hemorrhage (PBH) is a severe, poorly-prognostic stroke subtype. Since single biomarkers cannot reflect global inflammatory status, we constructed a composite Inflammatory Burden Index (IBI) incorporating neutrophil-to-lymphocyte ratio (NLR), C-reactive protein-to-albumin ratio (CAR), and systemic immune-inflammation index (SII), and assessed its links with 90-day poor functional outcome and stroke-associated pneumonia (SAP) in PBH patients.

Methods: 

This retrospective cohort included 347 PBH patients (2015–2023). The IBI was calculated as 0.50 × NLRz + 0.30 × CARz + 0.20 × SIIz, where the weights were assigned a priori from a qualitative synthesis of prior literature and expert judgment rather than derived from the study data. Patients were stratified into tertiles. Primary outcome was poor 90-day functional outcome (modified Rankin Scale 3–6). Secondary outcomes were 90-day mortality and SAP. Multivariable logistic regression, restricted cubic spline analysis, and decision curve analysis were performed. Incremental value over a reference clinical model (age, Glasgow Coma Scale, hematoma volume, intraventricular hemorrhage) was assessed via DeLong test, net reclassification improvement (NRI), and integrated discrimination improvement (IDI).

Results: 

The overall rates of poor outcome, mortality, and SAP were 30.3, 12.1, and 16.7%, respectively. Higher IBI tertiles were associated with progressively higher rates of poor functional outcome (18.1, 27.8, and 44.8%; p < 0.001), with graded but non-significant trends for SAP and mortality. In multivariable analysis, IBI independently predicted poor outcome (OR 2.31, 95% CI 1.64–3.26, p < 0.001) and SAP (OR 1.79, 95% CI 1.23–2.59, p = 0.002). RCS confirmed a monotonic dose–response relationship for poor outcome. IBI had the best AUC for poor outcome (0.663), outperforming SII and CAR, though not significantly better than NLR. Adding IBI to the clinical model improved AUC from 0.623 to 0.701 (DeLong p = 0.008; NRI 0.31, IDI 0.045). For SAP, NLR alone was marginally superior (AUC 0.615 vs. 0.614). Decision curve analysis showed net clinical benefit for the IBI model.

Conclusion: 

IBI independently predicts poor functional outcome and adds incremental value over clinical factors and most individual markers. However, its utility for SAP does not exceed that of NLR alone; hence IBI should be used selectively for functional prognostication, while NLR remains the preferred biomarker for SAP risk assessment, pending further external validation.

Graphical Abstract

Domain-specific functional outcome prediction in stroke rehabilitation: A multicenter artificial intelligence study

 Tell me PRECISELY HOW THIS GETS SURVIVORS RECOVERED!

Or just shut your fucking yaps and let some people with brains solve stroke! Somehow you are so blitheringly stupid you don't know predictions are completely fucking useless! WOW! IMPRESSIVE STUPIDITY!

Domain-specific functional outcome prediction in stroke rehabilitation: A multicenter artificial intelligence study

Author links open overlay panel, , , , , , , , , , 

Highlights

• •
  AI-based models predicted domain-specific functional outcomes after stroke, including ambulation, cognition, and ADLs.
• •
  Alignment-based regularization improved cross-institutional generalizability across four multicenter rehabilitation cohorts.
• •
  Domain-specific prediction models achieved AUROCs up to 0.897 in internal and 0.924 in external validation.
• •
  A prototype web-based decision support tool provides patient-specific recovery trajectories at 3 and 6 months after stroke.

Abstract

Background

Stroke is a leading cause of long-term disability worldwide, yet existing clinical decision support tools rely on global disability metrics, such as the modified Rankin Scale, which do not adequately reflect patient-centered rehabilitation or recovery goals.

Objective

We aimed to develop, validate, and implement artificial intelligence (AI)–based clinical support models for predicting domain-specific functional outcomes after stroke across multiple rehabilitation institutions and timepoints.

Methods

We utilized prospective and retrospective multicenter data collected from patients with stroke across four rehabilitation institutions in South Korea (2017–2024). Prognostic models were developed for three functional domains—ambulation, cognitive function, and activities of daily living—under two temporal scenarios: acute-to-subacute and acute-to-early-chronic prediction. Alignment-based regularization was applied to improve cross-institutional generalizability.

Results

The proposed framework achieved strong predictive performance in internal validation (AUROC up to 0.897 for ambulation and 0.864 for cognition) and favorable external validation performance, with AUROCs up to 0.924 (ADLs) and 0.892 (cognition), despite institution-specific differences in cohort size and variable availability across centers. The implemented web-based clinical decision support system provides real-time prediction of individualized recovery trajectory with intuitive visualization designed to support clinician–patient communication.

Conclusions

Our findings demonstrate the predictive feasibility of AI-based modeling for domain-specific stroke prognosis and present a prototype implementation illustrating the potential clinical applicability of the proposed framework across heterogeneous institutional settings. The use of routinely collected clinical variables and the preliminary cross-institutional validation results support further investigation of real-world rehabilitation practices, pending prospective clinical evaluation.

Quantitative multi-slice spiral CT perfusion parameters as predictors of collateral status and 90-day functional outcome in acute ischemic stroke

 

Predicting failure to recover IS STUPIDER THAN HELL! Deliver recovery you blithering idiots!

Send me personal hate mail on this: oc1dean@gmail.com. I'll print your complete statement with your name and title(If you can't stand by your name don't bother replying anonymously) and my response in my blog. Or are you afraid to engage with my stroke-addled mind? No excuses are allowed! You're medically trained; it should be simple to precisely state EXACTLY WHERE I'M WRONG.

Exactly what in this research gets survivors recovered? 100% recovery is the only goal in stroke; NOT PREDICTIONS, BIOMARKERS, PROGNOSTICATION, OR ASSESSMENTS! I'd fire anyone doing these!

Quantitative multi-slice spiral CT perfusion parameters as predictors of collateral status and 90-day functional outcome in acute ischemic stroke

• D

  Dan Zhu

• X

  Xiaozhou Ma

• Y

  Yingzhi Jiao

• S

  Shuai Liu

• J

  Jinzhu Yan

• 

  Lixin Zhang *

• Imaging Center, Qianwei Hospital of Jilin Province, Changchun, Jilin, China

Abstract

Objective: 

To evaluate the utility of quantitative computed tomography perfusion parameters for assessing collateral circulation and predicting 90-day functional outcomes in acute ischemic stroke.

Methods: 

This retrospective study included 82 patients who underwent perfusion imaging within 24 h of symptom onset. Parameters including relative cerebral blood flow, relative cerebral blood volume, mean transit time, time to maximum, and hypoperfusion intensity ratio were analyzed. Collateral status was classified using multiphase angiography, and 90-day outcomes were assessed using the modified Rankin Scale.

Results: 

Patients with robust collateral circulation showed higher relative cerebral blood flow and volume and shorter perfusion times compared with those with poor collaterals. Hypoperfusion intensity ratio demonstrated strong diagnostic performance (area under the curve 0.925). For predicting unfavorable 90-day outcome, HIR achieved an AUC of 0.912 and showed greater discrimination than mismatch ratio in the present cohort, while the combined HIR–rCBF–Tmax model achieved an AUC of 0.938. Hypoperfusion intensity ratio correlated positively with functional disability, while relative cerebral blood flow correlated negatively with infarct volume. Favorable outcomes were more frequent in patients with robust collaterals.

Conclusion: 

Quantitative CTP parameters bridge a natomical collateral assessment and downstream tissue-level perfusion. HIR may provide a particularly informative functional marker of collateral efficiency and 90-day prognosis beyond conventional mismatch assessment.

Platelet to high-density lipoprotein cholesterol ratio predicts clinical outcomes after acute ischemic stroke: a prospective cohort study

 

Predicting failure to recover IS STUPIDER THAN HELL! Deliver recovery you blithering idiots!

Send me personal hate mail on this: oc1dean@gmail.com. I'll print your complete statement with your name and title(If you can't stand by your name don't bother replying anonymously) and my response in my blog. Or are you afraid to engage with my stroke-addled mind? No excuses are allowed! You're medically trained; it should be simple to precisely state EXACTLY WHERE I'M WRONG.

Exactly what in this research gets survivors recovered? 100% recovery is the only goal in stroke; NOT PREDICTIONS, BIOMARKERS, PROGNOSTICATION, OR ASSESSMENTS! I'd fire anyone doing these!

Platelet to high-density lipoprotein cholesterol ratio predicts clinical outcomes after acute ischemic stroke: a prospective cohort study

• 

  Xuan Sun 1†

• H

  Haochen Sun 2†

• Z

  Zhijia Tang 3

• 

  Xinyang Qi 2

• X

  Xian Wang 4

• X

  Xiaoyin Wang 2*

• 1. Department of Clinical Laboratory, The Affiliated Brain Hospital of Nanjing Medical University, Nanjing, China

• 2. Department of Neurology, The Affiliated Brain Hospital of Nanjing Medical University, Nanjing, China

Abstract

Background: 

The platelet/high-density lipoprotein cholesterol ratio (PHR), a marker of hypercoagulable states and disordered lipid metabolism, has been confirmed as a predictor of cardiovascular disease. However, the effects of PHR on the prognosis of acute ischemic stroke (AIS) remain unknown. We aimed to assess the associations of PHR with the risk of clinical outcomes in patients with AIS.

Methods: 

This prospective observational study included 820 patients (median age, 68 years; female, 34.6%; median NIHSS at admission, 3) with AIS. The median time from symptom onset to admission was 2 days (interquartile range [IQR], 0–4), and from admission to blood sampling was 15 h (IQR, 12–19). PHR was calculated as platelet count (PC; 109 cells/L)/HDL-C (mmol/L) at admission. PHR was analyzed both as a continuous variable and in tertile form (tertile 1-tertile 3). To analyze the associations between PHR and clinical outcomes including all-cause death, stroke recurrence and poor functional outcome at 3 months, 6 months and 1 year, we used multivariable Cox and logistic regression, Kaplan–Meier survival curves, restricted cubic splines, subgroup analysis, concordance statistic (C-statistic), net reclassification index (NRI), and integrated discrimination improvement index (IDI).

Results: 

The median PHR was 202.155 (IQR, 153.120–262.365). Kaplan–Meier survival curves identified tertile 3 as the group with the highest risk for all-cause death and stroke recurrence. After adjustment, multivariable Cox regression (tertile 1 as reference) showed that the highest PHR tertile 3 was associated with increased risk for both all-cause death and stroke recurrence across all three follow-up intervals (3 months, 6 months and 1 year). In parallel, multivariable logistic regression (tertile 1 as reference) showed that tertile 3 was associated with a greater likelihood of poor functional outcome across the same three time points. Continuous PHR showed a positive dose–response relationship with clinical outcomes. Subgroup analysis revealed significant interactions of age (p < 0.05) with PHR for all-cause death, and of BMI (p < 0.05) with PHR for mRS 3–6. A basic model’s predictive ability was strengthened by the addition of PHR (C-statistic, NRI, IDI).

Conclusion: 

A higher PHR level in patients with AIS is strongly associated with an increased risk of all-cause death, stroke recurrence and poor functional outcome. As a valuable predictive biomarker, PHR may provide a simple and effective tool for predicting clinical outcomes in patients with AIS.

Graphical Abstract