Speech Analysis App Predicts Worsening Heart Failure Before Symptom Onset

A voice analysis app used by heart failure patients at home recognises fluid in the lungs three weeks before an unplanned hospitalisation or escalation in outpatient drug treatment. The late breaking research is presented at Heart Failure 2022, a scientific congress of the European Society of Cardiology (ESC).(1)

"The current standard of care just isn’t good enough for keeping patients with heart failure well and out of the hospital," said study author Professor William Abraham of The Ohio State University, Columbus, US. "The system tested in this study was able to predict 80% of worsening heart failure in advance, compared to a 10-20% success rate for daily weight monitoring shown in previous studies. In the future, speech analysis, together with other clinical information, could be used to modify treatments before a patient’s condition deteriorates and thereby avoid hospital admission."

In patients with heart failure, the heart does not pump blood around the body as well as it should. Fluid is not eliminated properly by the kidneys and this excess fluid builds up in the lungs or legs. Lung congestion is a common cause of hospitalisation, and can be life-threatening. Currently, lung congestion is monitored by asking patients to weigh themselves every day and report any substantial gain. Patients are also requested to report worsening symptoms such as shortness of breath, needing to elevate the head at night to breath comfortably and sleep, and swelling in the feet or ankles.

The speech analysis app used in the current study was previously shown to detect fluid in the lungs of patients hospitalised with acute heart failure.(2) This study investigated its ability to predict worsening heart failure in patients living at home.

The study included 180 patients with heart failure taking guideline recommended medications. At the beginning of the study, as a baseline measurement, participants recorded five sentences on a standard smartphone using the voice analysis app. During the study period, every morning before breakfast patients recorded the same five sentences using the app. The app compared each day's recordings with the baseline versions and alerted research staff when it detected lung congestion. Professor Abraham explained: "The system sets a baseline for each individual patient during a period of stability. It then detects changes in speech over time that indicate fluid in the lungs."

The researchers examined whether the lung congestion alerts predicted heart failure events, defined as at least one worsening symptom which led to hospitalisation or escalation in outpatient drug treatment. They did this by comparing the date of the alert with the dates of subsequent heart failure events. "True alerts" were those issued within a 31-day window before the heart failure event. "False alerts" were those issued outside of the 31-day window prior to an event.

Professor Abraham noted that a "true alert" would continue to be issued daily until the heart failure event occurred. He said: "In this study, the alerts were not acted upon by clinicians since the aim was to determine the diagnostic performance of the tool. In real life, notifications would be sent to the patient's physician until it was responded to. Here we followed patients longitudinally for worsening heart failure events and then looked to see if there was an alert prior to that event - i.e. a 'true alert'. If there was no alert then the system failed to predict the event. Alerts not followed by a worsening heart failure event were 'false alerts'."

During the study, patients provided recordings for an average of 512 days. A total of 49 heart failure events occurred in 37 patients, of which 39 (80%) were correctly predicted by the app and 10 (20%) were missed. True alerts were issued a median of 21 days before worsening symptoms. Each patient received a false alert every 4.8 months on average, resulting in an average of 2.5 erroneous warnings each year.

Professor Abraham said: "In this community-based study, a voice analysis app was able to predict most cases of worsening heart failure well in advance, with very few false alarms. Weight gain and symptoms occur too late to allow medical interventions that keep patients out of the hospital. Future studies will investigate whether changing patient management following an alert, for example by increasing the diuretic dose to get rid of excess fluid, can prevent hospitalisations."

1. The abstract 'HearO™ community study: remote speech analysis to predict worsening heart failure events' has been presented during the session 'Late-Breaking Science - Devices' on 21 May at 15:50 CEST in Room 3.
2. Amir O, Abraham WT, Azzam ZS, et al. Remote speech analysis in the evaluation of hospitalized patients with acute decompensated heart failure. JACC Heart Fail. 2022;10:41-49.

Most Popular Now

AI-Powered CRISPR could Lead to Faster G…

Stanford Medicine researchers have developed an artificial intelligence (AI) tool to help scientists better plan gene-editing experiments. The technology, CRISPR-GPT, acts as a gene-editing “copilot” supported by AI to help...

Groundbreaking AI Aims to Speed Lifesavi…

To solve a problem, we have to see it clearly. Whether it’s an infection by a novel virus or memory-stealing plaques forming in the brains of Alzheimer’s patients, visualizing disease processes...

AI Spots Hidden Signs of Depression in S…

Depression is one of the most common mental health challenges, but its early signs are often overlooked. It is often linked to reduced facial expressivity. However, whether mild depression or...

ChatGPT 4o Therapeutic Chatbot 'Ama…

One of the first randomized controlled trials assessing the effectiveness of a large language model (LLM) chatbot 'Amanda' for relationship support shows that a single session of chatbot therapy...

AI Tools Help Predict Severe Asthma Risk…

Mayo Clinic researchers have developed artificial intelligence (AI) tools that help identify which children with asthma face the highest risk of serious asthma exacerbation and acute respiratory infections. The study...

AI Model Forecasts Disease Risk Decades …

Imagine a future where your medical history could help predict what health conditions you might face in the next two decades. Researchers have developed a generative AI model that uses...

AI Distinguishes Glioblastoma from Look-…

A Harvard Medical School–led research team has developed an AI tool that can reliably tell apart two look-alike cancers found in the brain but with different origins, behaviors, and treatments. The...

AI Model Indicates Four out of Ten Breas…

A project at Lund University in Sweden has trained an AI model to identify breast cancer patients who could be spared from axillary surgery. The model analyses previously unutilised information...

Smart Device Uses AI and Bioelectronics …

As a wound heals, it goes through several stages: clotting to stop bleeding, immune system response, scabbing, and scarring. A wearable device called "a-Heal," designed by engineers at the University...

Overcoming the AI Applicability Crisis a…

Opinion Article by Harry Lykostratis, Chief Executive, Open Medical. The government’s 10 Year Health Plan makes a lot of the potential of AI-software to support clinical decision making, improve productivity, and...

Dartford and Gravesham Implements Clinis…

Dartford and Gravesham NHS Trust has taken a significant step towards a more digital future by rolling out electronic test ordering using Clinisys ICE. The trust deployed the order communications...