Talk:Lentis/Ellie, the Microsoft Kinect, and Psychotherapy

Diagnosis in Psychotherapy Clinical diagnosis of mental disorders has evolved significantly over the last century. In the early 1900s, professionals in the nascent fields of psychology and psychiatry focused primarily on describing symptoms of mental illnesses which they observed. Sigmund Freud, in the 1920s and 30s, shifted the field by investigating causes of mental disorders; however, psychotherapists from this era tended to attribute mental illnesses to intentional misbehavior or personal weaknesses. It was only in 1980, when DSMIII, the third edition of the Diagnostic and Statistical Manual of Mental Disorders was published, that mental disorders began to be seen as legitimate medical conditions which could be diagnosed with an evidence-based scientific system. The Rise of Computer Aids in Psychological Diagnosis Joseph Weizenbaum and a team at MIT led the first documented attempt to create a computerized psychologist in 1966. Their system, known as ELIZA, was a computer program which could somewhat-intelligently generate responses to human questions or comments in order to “chat” with a real person. Though the system was not very robust and did not actually add much value in terms of psychological testing, it was a significant advancement in artificial intelligence and presented the concept of a “virtual human” medical professional. Other computer programs, such as Gibbons et al.’s (2013) computerized adaptive test or Spitzer & Endicott’s (1968) DIAGNO, have been developed to provide algorithms for efficiently gathering patient information. These systems work through a tree structure of diagnostic questions, using a patient’s previous answers to determine relevant questions to ask in order to narrow down the list of possible disorders. More and more psychologists are recognizing the risk of biases leading them down the wrong diagnostic path. Anchoring bias – when a clinician comes up with a preliminary diagnosis and then continues to attribute other observed symptoms to that diagnosis even if they do not quite fit – is a common risk. Ellie and the SimSensei system attempts to fight these biases by combining a virtual human interface with the ability to gather and analyze large amounts of patient behavioral data. Ellie and the SimSensei Project Ellie was developed by computer scientist Louis-Philippe Morency and psychologist Albert “Skip” Rizzo, along with other staff at the University of Southern California’s Institute of Creative Technologies (ICT). The SimSensei project, which received funding from the Defense Advanced Research Projects Agency (DARPA), aims to create a platform to help diagnose PTSD, anxiety, and depression in military personnel and their families. The U.S. Department of Veterans Affairs estimates that PTSD affects a significant portion of returning military members, including up to 31 percent of Vietnam veterans. Both current and retired military members may feel stigmatized for seeking professional help in dealing with these disorders, which is where Ellie comes in.

[after Sarah’s part on changing perceptions of PTSD] '''Shift toward objective measures in psychological diagnosis '''[Sarah? Or what I wrote before?] '''Support for and opposition of computer-aided diagnosis '''Several studies have demonstrated that psychological diagnosis can be unreliable, and attribute inconsistency in diagnoses to numerous factors, such as a clinician’s bias toward present or the most pressing symptoms, inability to make the patient feel comfortable enough to provide truthful and complete information, or inability to conduct a comprehensive examination due to a large workload and limited time (Reliability of psych diagnosis sources). In addition to clinician errors, the psychiatric classification system can be to blame – Ward et al. found that 62.5% of the cases of diagnostic disagreement between psychiatrists that they studied could be attributed to the lack of a precise diagnostic term. The creators and funders of Ellie and other computer-aided diagnosis systems, such as IBM’s Watson and Isabel Healthcare’s Symptom Checker, believe that their systems will improve the reliability of both psychological and medical diagnosis by allowing a more comprehensive assessment to check more information gathered about a patient from a larger set of medical knowledge than a single clinician can learn. According to Sloan-Kettering Memorial Hospital, where IBM’s Watson is being trained, doctors would need to read at least 160 hours per week in order to stay up to date on newly published medical knowledge. Sloan-Kettering also reports that the diagnoses and treatment decisions made by human doctors are typically based on a set of information, only 20% of which comes from trial-based evidence (CITE Wired article source). Systems such as Ellie could reduce the impact of clinicians’ biases when making diagnoses. Psychologists have found that when the same core set of information is presented in different ways to two study groups, the groups come to different decisions (Cite atlantic article source). Other experts warn of anchoring bias – the phenomenon where clinicians tend to use additional patient information they receive to justify their initial diagnosis rather than reevaluating the complete set of data (cite). However, there is also opposition to computerized diagnostic aids. Many of the concerns voiced about the general rise of automation can be applied to the case of Ellie. Jan Noyes, an ergonomics expert at the University of Bristol, argues that the increased reliance on autopilot systems can “de-skill” flight crew, so that when the computer system fails, they are not able to respond (cite Atlantic article). Within the medical field, similar arguments were made against the use of x-ray imaging in the decades after it was invented in 1895. While most heralded it as a method that would revolutionize and improve surgical practice, a Yale professor of medicine worried that doctors would blindly use and become too reliant on x-ray images and discount the importance of human judgment that should complement the data. “The worst problems come when the doctor fits the patients to his skills…we must get back to training students to look at the patient rather than simply the data base,” he stated (cite book). Because Ellie has only been tested with a few patient groups and is one of the first computer systems to aid psychological diagnosis (check if this is true?), supporting and opposing views of the system have not fully emerged.

White collar job replacement'Bold text' The rise of robotics and automation has often been linked to the decline in manufacturing jobs and other blue collar work in recent decades; however, systems like Ellie have been raising concerns that computers could also be encroaching on white collar jobs. While at the very least human psychiatrists and doctors will still be needed to assess the data collected by Ellie, Watson, Isabel, and other computer-aided diagnostic systems to make a diagnosis, these systems will significantly reduce the amount of human time needed to screen patients, which could lead to a need for a fewer number of medical professionals. In other fields, computer programs have been able to write news stories when given data (Narrative Science) and screen through thousands of legal documents in place of lawyers to select the ones most relevant to a case (cite article). The Pew Research Center interviewed X experts in ___ and found that there is general consensus that some white collar jobs will be replaced by computers over the next decade; however, they are deeply divided about whether or not new types of jobs will be created to offset this change.