Summary: A novel AI tool identified long COVID in 22.8 % of people, a much higher rate than originally diagnosed. By analyzing intensive health records from virtually 300, 000 patients, the engine identifies huge COVID by distinguishing symptoms linked especially to SARS-CoV-2 disease rather than pre-existing conditions.
This “precision phenotyping” AI technique, which helps clinicians differentiate long COVID symptoms from other health conditions, may increase diagnostic accuracy by about 3 %.
Important Information:
- AI-based detail phenotyping: Recognizes long COVID just after excluding different causes of symptoms in health records, improving medical reliability.
- Broader picture: Algorithm symptoms mirror the Massachusetts demographic report, addressing prejudices found in traditional medical codes.
- Potential for research: Long COVID subtypes ‘ biological and metabolic factors may be improved by algorithms.
Origin: Harvard
While earlier clinical studies have suggested that 7 percent of the population suffers from long COVID, a fresh AI tool developed by  , Mass General Brigham , revealed a little higher 22.8 percentage, according to the research.  ,
The AI-based device can assist professionals detect long-term COVID cases by sifting through electronic health records. The often-mysterious state may contain a , myriad of , enduring symptoms,  , including fatigue, chronic coughing, and head cloud after infections from SARS-CoV-2.  ,
De-identification patient data was derived for the engine, which was developed by the Mass General Brigham system’s almost 300,000 individual records in 14 hospitals and 20 group health centers.
The , outcomes,  , published in the journal , MedRxiv, had determine more people who should be receiving treatment for this possibly debilitating situation.
Hossein Estiri, senior author and associate professor of medicine at Harvard Medical School, said,” Our AI tool may transform a tense diagnostic process into something strong and focused,” giving practitioners the ability to create sense of a difficult situation.  ,
” With this work, we may finally be able to see long COVID for what it truly is — and more importantly, how to treat it”.
For the purposes of their study, Estiri and colleagues defined long COVID as a , diagnosis of exclusion , that is also , infection-associated. That implies that the diagnosis was related to a COVID infection and could n’t be explained in the patient’s own medical history. In addition, a 12-month follow-up period required the diagnosis to have persisted for at least two months.  ,
The novel technique used by Estiri and colleagues, known as “precision phenotyping,” uses individual records to classify symptoms and conditions related to COVID-19 and to track them against other conditions over time.
For instance, the algorithm can determine whether shortness of breath is caused by pre-existing conditions like asthma or heart failure rather than long COVID. The tool would only indicate that the patient had long COVID when all other options had been exhausted.  ,
” Physicians are frequently faced with having to balance busy caseloads by wading through a tangled web of symptoms and medical histories. Having a tool powered by AI that can methodically do it for them could be a game-changer”, said , Alaleh Azhir, co-lead author and an , internal medicine resident at Brigham and Women’s Hospital, a founding member of the Mass General Brigham healthcare system.
Researchers noted diagnoses with the official ICD-10 diagnostic code for long COVID trend toward those with easier access to healthcare and the new tool’s patient-centered diagnoses may also help reduce biases built into current diagnostics for long COVID.
The researchers said their tool is about 3 percent more accurate than the data ICD-10 codes capture, while being less biased. In particular, their study demonstrated that long COVID, in contrast to long COVID algorithms that rely on a single diagnostic code or individual clinical encounters, skews results toward specific populations, such as those with greater access to care, in comparison to long COVID individuals.
” This broader scope ensures that marginalized communities, often sidelined in clinical studies, are no longer invisible”, said Estiri.
The study’s and AI tool’s limitations include the possibility that the algorithm’s health record data, which accounts for prolonged COVID symptoms, may not be as comprehensive as the information that doctors record following a clinical visit.
Another issue was that the algorithm failed to account for potential worsening of a previously severe COVID symptom. For instance, the algorithm might have removed the episodes even if they were long COVID indicators if a patient had COPD that had gotten worse before COVID-19 developed.
It’s also difficult to tell when a patient may have received COVID-19 after recent declines in COVID-19 testing.
The study was limited to patients in Massachusetts.
In larger cohorts of patients with particular illnesses like COPD or diabetes, the algorithm may be used. Additionally, the researchers intend to make this algorithm available for free public viewing so that doctors and healthcare systems can use it in their patient populations.  ,
This research may help lay the groundwork for future research into the genetic and biochemical factors that underlie the various subtypes of long COVID.
Estiri said,” Questions about the true burden of long COVID– questions that have so far remained elusive– now seem more in the way of reach.”
Funding: Support was given by the National Institutes of Health, National Institute of Allergy and Infectious Diseases ( NIAID ) R01AI165535, National Heart, Lung, and Blood Institute ( NHLBI ) OT2HL161847, and National Center for Advancing Translational Sciences (NCATS ) UL1 TR003167, UL1 TR001881, and U24TR004111.
The German Research Foundation ( 426671079 ), the German Academic Exchange Service ( DAAD), the Federal Ministry of Education and Research ( BMBF), and the German Academic Exchange Service ( DAAD ) provided some funding for J. Hügel’s research.
About this long-running COVID research story.
Author: MGB Communications
Source: Harvard
Contact: MGB Communications – Harvard
Image: The image is credited to Neuroscience News
Original Research: Open access.
” Precision Phenotyping for Curating Research Cohorts of Patients with Post-Acute Sequelae of COVID-19 ( PASC ) as a Diagnosis of Exclusion” by Hossein Estiri et al. MedRxiv
Abstract
Precision Phenotyping for Curating Research Cohorts of Patients with Post-Acute Sequelae of COVID-19 ( PASC ) as a Diagnosis of Exclusion
Scalable identification of patients with the post-acute sequelae of COVID-19 ( PASC ) is challenging due to a lack of reproducible precision phenotyping algorithms and the suboptimal accuracy, demographic biases, and underestimation of the PASC diagnosis code ( ICD-10 U09.9 ).
We created a precise phenotyping algorithm for identifying PASC patient research cohorts in a retrospective case-control study with an exclusion diagnosis. We analyzed longitudinal electronic health record ( EHR ) data from more than 295 000 patients at Massachusetts’s 14 hospitals and 20 community health centers.
To exclude sequelae that previous conditions can explain, the algorithm employs an attention mechanism. Our precise phenotyping algorithm was tested and validated by independent chart reviews.
Our PASC phenotyping algorithm improves accuracy and prevalence estimation, as well as reducing bias when identifying long COVID patients in comparison to the U09.9 diagnosis code.
Our algorithm generated a 79.9 % accuracy for a PASC research cohort of over 24 thousand patients ( compared to 67.8 % for the U09.9 diagnosis code ).
Our estimated prevalence of PASC was 22.8 percent, which is close to the national estimates for the region. Additionally, we provide a thorough analysis that includes identified lingering effects by organ, comorbidity profiles, and temporal differences in the risk of PASC.
The PASC phenotyping technique used in this study has better accuracy, accurate identification of PASC patients, and less bias when identifying Long COVID patients.
The PASC cohort derived from our algorithm will serve as a springboard for delving into Long COVID’s genetic, metabolomic, and clinical intricacies, surmounting the constraints of recent PASC cohort studies, which were hampered by their limited size and available outcome data.
