Researchers developed a prediction tool that can identify sepsis patients at high risk for PICS—a serious long-term condition involving ongoing inflammation and weakness—with about 80% accuracy using only blood tests from the patient’s first ICU day. According to Gram Research analysis, high lactate levels, low immune cell counts, and low albumin were the strongest predictors, with respiratory infections carrying the highest risk. The tool showed consistent results when tested on new patients, though it requires further testing in multiple hospitals before doctors can use it in regular practice.
Researchers created a prediction tool to identify sepsis patients at highest risk of developing a serious long-term condition called PICS (persistent inflammation-immunosuppression-catabolism syndrome). By analyzing blood tests and infection type from 242 patients in the ICU, they found that certain markers—like high lactate levels and low lymphocyte counts—could predict who would struggle with prolonged recovery. According to Gram Research analysis, this tool correctly identified at-risk patients about 80% of the time, potentially helping doctors plan better care for sepsis survivors who need extended ICU stays.
Key Statistics
A 2026 cohort study of 242 sepsis patients found that a prediction model using admission-day blood tests correctly identified patients at risk for PICS (persistent inflammation-immunosuppression-catabolism syndrome) with 80% accuracy in both the development and validation groups.
Among 242 sepsis patients requiring ICU stays of at least 14 days, 38% developed PICS, and those with respiratory infections were significantly more likely to develop the condition compared to patients with urinary, gastrointestinal, or skin infections.
High lactate levels increased the odds of developing PICS by 20% per unit increase, while higher lymphocyte counts, albumin levels, and vitamin D levels were each independently associated with substantially lower PICS risk in sepsis patients.
The prediction model achieved an area under the curve (AUC) of 0.803 in the original patient group and 0.791 in a new validation group, demonstrating consistent performance when tested on different sepsis patients.
The Quick Take
- What they studied: Can doctors predict which sepsis patients will develop PICS, a condition where the body stays inflamed and weakened for weeks or months after the initial infection?
- Who participated: 242 sepsis patients who stayed in the ICU for at least 2 weeks. About 38% of them developed PICS during their hospital stay.
- Key finding: A simple prediction model using blood tests taken on the first day in the ICU could correctly identify 80% of patients who would develop PICS. The model was tested twice and worked consistently both times.
- What it means for you: If you or a loved one develops sepsis requiring long ICU stays, doctors may soon be able to predict early on who needs extra monitoring and specialized care to prevent long-term complications. This is still being tested and not yet used in most hospitals.
The Research Details
Researchers looked at two groups of sepsis patients: one group from the past (January 2023 to May 2025) to develop their prediction tool, and another group from more recent months (June 2025 to February 2026) to test if the tool actually worked. This approach, called ‘ambispective,’ combines looking backward at past records with looking forward at new patients.
They used a statistical method called LASSO regression to identify which blood test results and infection types were most important for predicting PICS. Then they built a prediction model using standard statistical techniques. The model was like a scoring system that combined multiple factors to estimate each patient’s risk.
They measured how well the model worked by checking if it correctly identified high-risk patients and whether its predictions matched what actually happened. They also tested whether doctors would actually find it useful in real-world practice.
This research approach is important because it tests the prediction tool twice—once on the data used to create it, and again on completely new patients. This ‘validation’ step is crucial because prediction tools often work better on the data they were built from than on new patients. By testing it twice, researchers can be more confident the tool will actually help doctors in real hospitals.
The study has several strengths: it used a clear definition of PICS, measured important blood markers at a specific time point (ICU admission), and tested the model on new patients to verify it worked. However, it was conducted at only one hospital, so results may not apply everywhere. The researchers themselves note that larger, multi-center studies are needed before this tool should be used in regular clinical practice. The model’s accuracy (80%) is good but not perfect, meaning some patients will be misclassified.
What the Results Show
The prediction model correctly identified at-risk patients with an accuracy score of 0.803 in the original group and 0.791 in the new patient group. (A perfect score would be 1.0, and random guessing would be 0.5, so this performance is considered good.) This consistency between the two groups suggests the model is reliable.
The most important factors for predicting PICS were: high lactate levels in the blood (a sign the body isn’t getting enough oxygen), low lymphocyte counts (a type of immune cell), low albumin (a protein that shows nutrition status), and low vitamin D levels. Patients with these markers were at higher risk of developing PICS.
The type of infection also mattered significantly. Patients with respiratory infections (lung infections) were at highest risk for PICS. Those with urinary tract infections, stomach/intestinal infections, or skin infections had much lower risk—up to 93% lower for urinary infections compared to respiratory infections.
Among the 242 patients studied, 93 (about 38%) developed PICS. The model successfully predicted most of these cases using only information available on the patient’s first day in the ICU.
The research identified specific blood markers that doctors could easily measure: lactate, lymphocyte count, albumin, and vitamin D. These are routine tests already done in most ICUs, making the prediction model practical to implement. The finding that infection type matters—with respiratory infections being riskier—could help doctors prioritize monitoring for certain patients. The model’s ability to work with admission-day data is significant because it means doctors could identify high-risk patients immediately, not weeks into their hospital stay.
PICS is a relatively newly recognized condition in sepsis patients, so there are limited previous prediction models to compare this to. This study appears to be one of the first to create and validate a practical prediction tool using only admission-day variables. Previous research has identified individual risk factors for PICS, but this study combines them into a single, usable tool. The accuracy rates (around 80%) are comparable to or better than prediction models for other serious ICU conditions.
The study was conducted at only one hospital in one country, so results may not apply to all populations or healthcare settings. The model was built and tested on a relatively small number of patients (242 total), which is adequate but not large. The researchers did not test whether using this prediction tool actually changed patient outcomes or improved care—they only showed it could predict who would get PICS. The study also didn’t explore whether the model works equally well for different age groups, genders, or types of sepsis. Finally, the model needs testing in multiple other hospitals before doctors should use it routinely in clinical practice.
The Bottom Line
This prediction model shows promise but is not yet ready for routine clinical use. Confidence level: MODERATE. Doctors should be aware this tool exists and may help identify high-risk sepsis patients in the future. For now, it should only be used in research settings or clinical trials. Patients with sepsis who stay in the ICU for extended periods should ensure their doctors monitor key markers like lactate, lymphocyte count, albumin, and vitamin D, as these appear important for predicting complications.
This research is most relevant to: (1) ICU doctors and nurses caring for sepsis patients, (2) sepsis survivors and their families who want to understand long-term risks, (3) hospital administrators planning ICU resources, and (4) researchers developing better sepsis care protocols. It’s less immediately relevant to people without sepsis or those with mild infections. Patients with respiratory infections requiring ICU admission should be particularly aware of higher PICS risk.
PICS typically develops over weeks to months in hospitalized sepsis patients. If this prediction tool were used clinically, doctors could identify high-risk patients within the first day of ICU admission. However, seeing actual improvements in patient outcomes would require months to years of follow-up care. The tool itself would provide information immediately, but benefits would unfold over the patient’s entire recovery period.
Frequently Asked Questions
What is PICS and why do some sepsis patients develop it?
PICS is persistent inflammation-immunosuppression-catabolism syndrome, a condition where sepsis survivors remain weak, inflamed, and vulnerable to infections for weeks or months. About 38% of sepsis patients requiring extended ICU stays develop PICS. The exact cause isn’t fully understood, but low immune cell counts, poor nutrition, and high lactate levels appear to increase risk.
Can doctors predict who will get PICS after sepsis?
A new prediction tool using blood tests from the first ICU day can identify high-risk patients with about 80% accuracy. The tool measures lactate, lymphocyte count, albumin, and vitamin D levels. However, this tool is still being tested and isn’t yet used in most hospitals for routine clinical decisions.
Which sepsis patients are most likely to develop PICS?
Patients with respiratory (lung) infections are at highest risk for PICS. Those with urinary tract infections, stomach infections, or skin infections have much lower risk—up to 93% lower for urinary infections. Patients with high lactate, low immune cells, low albumin, and low vitamin D are also at greater risk.
What blood markers predict PICS in sepsis patients?
Four key markers measured on ICU admission predict PICS risk: high lactate (indicates poor oxygen delivery), low lymphocyte count (weak immune system), low albumin (poor nutrition), and low vitamin D (nutritional deficiency). These are routine tests already done in most ICUs, making screening practical.
When will doctors use this PICS prediction tool in hospitals?
The tool shows promise but requires additional testing in multiple hospitals before routine clinical use. Researchers recommend external validation studies and evaluation of whether using the tool actually improves patient outcomes. This process typically takes 1-3 years, so widespread adoption is likely several years away.
Want to Apply This Research?
- For sepsis survivors: Track weekly blood work results (lactate, lymphocyte count, albumin, vitamin D levels) for the first 3 months after ICU discharge. Note any persistent symptoms like fatigue, weakness, or frequent infections. Create a simple chart showing whether these markers are improving or staying low.
- If you’re a sepsis survivor, work with your doctor to ensure you’re getting adequate nutrition (supporting albumin levels), vitamin D supplementation if deficient, and monitoring for signs of ongoing infection or immune problems. Request regular blood work to track these specific markers during recovery.
- Set up monthly check-ins with your healthcare provider for the first 6 months post-sepsis to monitor recovery progress. Use the app to log symptoms, blood test results, and functional improvements (like returning to normal activities). Share this data with your doctor to help identify if you’re developing PICS-related complications early.
This research describes a prediction tool still in development and not yet approved for routine clinical use. It should not be used to make medical decisions without consultation with a healthcare provider. If you or a loved one has sepsis or is recovering from sepsis, discuss all treatment and monitoring decisions with your ICU team or primary care doctor. This article summarizes research findings and does not constitute medical advice. Always consult qualified healthcare professionals for diagnosis, treatment, and management of sepsis or any medical condition.
This research translation is published by Gram Research, the science division of Gram, an AI-powered nutrition tracking app.
