|
In our continuing saga to become isolated from the rest of the world, researchers are developing software that can analyze a person’s speech patterns to tell how they are feeling.
Your Computer May Soon Be Your Therapist
People often complain nowadays that computers are replacing personal interactions. Well, if MIT’s Media Lab succeeds in one of their latest development projects, computers may eventually replace therapists.
Professor Sandy Pentland, a researcher with the MIT Media Lab, has a group of researchers working to develop algorithms in speech recognition software that can help determine whether a person is feeling anxious, awkward, disconnected from others, or even depressed. Most speech recognition software is focused on turning phrases and words into written text, but Pentlant’s scientists are fine-tuning the algorithms to analyze very subtle cues within a person’s speech patterns that can reveal a lot about their moods and feelings. Cogito Health of Charlestown MA, is employing Pentland’s research in their voice-analysis software program that screens telephone callers for depression.
Psychiatrists have identified a characteristic pattern recognized in the speech of many people diagnosed with clinical depression. They speak slowly, haltingly, in a quiet monotone. Joshua Feast, Cogito’s CEO, is working with his colleagues to train computers to recognize these vocal patterns in recorded audio samples. Feast says that the software, after it is perfected, could be a valuable tool for doctors and psychiatrists to use when managing patients dealing with chronic disease, which often leads to despair and depression. Read more on how does speech recognition work.
In many disease-management programs, specially trained nurses call patients regularly to check on them, see if they are taking their medication, and be available to answer questions the patients may have. But symptoms of depression are not easy to identify over the telephone, and nurses are not always trained to recognize them in speech patterns. Feast claims that voice analysis software that detects moods would provide a noninvasive method for nurses to be able to screen for depression when they make their routine checkup calls. "If you’re a nurse and you’re trying to deal with a patient with long-term diabetes, it's very hard to tell if a person is depressed," says Feast. "We try to help nurses detect possible mood disorders in patients that have chronic disease."
Pharmaceutical giant Pfizer had a similar idea recently when it developed a voice analysis software program that could detect the early signs of Parkinson’s disease. Scientists designed the software specifically to identify tiny tremors within a patient’s speech patterns, which were able to offer clues to help doctors gauge their physical responses to certain medications.
Cogito Health is working to refine their software to detect specific patterns in audio recordings to measure the consistency of a speaker in tone, energy, fluidity, and engagement in the conversation - such as whether a person responds with a simple acknowledgment or even silence, rather than responding by talking. Pentland explains that the software doesn’t listen to the words, but rather to the patterns within the speech. By measuring the signals contained in the speech, doctors can tell what is going on in the patient’s mind. Such technology can help doctors monitor the long-term progress of patients, rather than seeing them only every few weeks or months.
Cogito is conducting a large-scale software trial using a collection of hundreds of telephone conversations between patients and nurses. After giving the patients follow-up questions to determine which of the patients were depressed, researchers used the software to evaluate the audio recordings to see if it was able to identify the depressed patients accurately. Cogito plans to design software that can recognize other mood disorders within at-risk groups of people, such as post-traumatic stress affecting soldiers. Being able to detect psychological despair early and provide treatment can make an enormous difference in the outcome of post-traumatic situations.
The trials are still ongoing, but so far Feast says the results are encouraging. First results are expected to be published in 2010
Your Computer May Soon Be Your Therapist
People often complain nowadays that computers are replacing personal interactions. Well, if MIT’s Media Lab succeeds in one of their latest development projects, computers may eventually replace therapists.
Professor Sandy Pentland, a researcher with the MIT Media Lab, has a group of researchers working to develop algorithms in speech recognition software that can help determine whether a person is feeling anxious, awkward, disconnected from others, or even depressed. Most speech recognition software is focused on turning phrases and words into written text, but Pentlant’s scientists are fine-tuning the algorithms to analyze very subtle cues within a person’s speech patterns that can reveal a lot about their moods and feelings. Cogito Health of Charlestown MA, is employing Pentland’s research in their voice-analysis software program that screens telephone callers for depression.
Psychiatrists have identified a characteristic pattern recognized in the speech of many people diagnosed with clinical depression. They speak slowly, haltingly, in a quiet monotone. Joshua Feast, Cogito’s CEO, is working with his colleagues to train computers to recognize these vocal patterns in recorded audio samples. Feast says that the software, after it is perfected, could be a valuable tool for doctors and psychiatrists to use when managing patients dealing with chronic disease, which often leads to despair and depression. Read more on how does speech recognition work.
In many disease-management programs, specially trained nurses call patients regularly to check on them, see if they are taking their medication, and be available to answer questions the patients may have. But symptoms of depression are not easy to identify over the telephone, and nurses are not always trained to recognize them in speech patterns. Feast claims that voice analysis software that detects moods would provide a noninvasive method for nurses to be able to screen for depression when they make their routine checkup calls. "If you’re a nurse and you’re trying to deal with a patient with long-term diabetes, it's very hard to tell if a person is depressed," says Feast. "We try to help nurses detect possible mood disorders in patients that have chronic disease."
Pharmaceutical giant Pfizer had a similar idea recently when it developed a voice analysis software program that could detect the early signs of Parkinson’s disease. Scientists designed the software specifically to identify tiny tremors within a patient’s speech patterns, which were able to offer clues to help doctors gauge their physical responses to certain medications.
Cogito Health is working to refine their software to detect specific patterns in audio recordings to measure the consistency of a speaker in tone, energy, fluidity, and engagement in the conversation - such as whether a person responds with a simple acknowledgment or even silence, rather than responding by talking. Pentland explains that the software doesn’t listen to the words, but rather to the patterns within the speech. By measuring the signals contained in the speech, doctors can tell what is going on in the patient’s mind. Such technology can help doctors monitor the long-term progress of patients, rather than seeing them only every few weeks or months.
Cogito is conducting a large-scale software trial using a collection of hundreds of telephone conversations between patients and nurses. After giving the patients follow-up questions to determine which of the patients were depressed, researchers used the software to evaluate the audio recordings to see if it was able to identify the depressed patients accurately. Cogito plans to design software that can recognize other mood disorders within at-risk groups of people, such as post-traumatic stress affecting soldiers. Being able to detect psychological despair early and provide treatment can make an enormous difference in the outcome of post-traumatic situations.
The trials are still ongoing, but so far Feast says the results are encouraging. First results are expected to be published in 2010
