Big data has become a buzzword in the world of commerce, healthcare, and technology. A field that has only recently started to wade into the unchartered waters of data science is mental health. Big data could be beneficial in mental health at three phases of treatment:
- Pre-diagnosis (predictive)
To take advantage of the benefits of data analytics, we would firstly need to build a large data warehouse or a ‘mental health data universe’. This means gathering data from a multitude of sources, linking patient data together, and storing it in one central location. With a large enough data set, analysis and training could then be performed on the data to generate psychological profiles of mental health conditions, illnesses, and disorders. When new patient cases are presented to clinicians, psychologists, or therapists, the details of their case can be matched with one of the existing data profiles, thereby giving deeper insights into the patients state of mind.
The Importance of Privacy and Ethics
When handling mental health data, there needs to be provision to protect patient privacy and ensure consent requirements, data protection, and other ethical concerns are met. For this, new approaches and laws may be required. We cannot hope to achieve the benefits of such important data without at the same time securing sensitive information from prying eyes and machines. Particularly if those prying eyes have malicious or selfish intent.
One example of the potential misuse of data and AI involves Facebook. In March 2017, Facebook launched a project to prevent suicide using its data. Then, two months later, a leaked confidential document revealed that the company had shared collated data about teens feeling ‘insecure’ or ‘worthless’ with a potential advertiser. Although this example is not about the misuse of clinical mental health data, it certainly raises concerns over the lack of privacy laws for global data controllers like Facebook.
How can we use the data for psychological profile matching?
Profile matching could be performed during any of the three previously mentioned phases of treatment in the following ways:
- Pre-diagnosis: flag individuals who may be likely to develop a mental illness in the short or long-term future
- Diagnosis: determine the diagnosis for a patient currently seeking therapy
- Therapy: generate treatment plans and choose the therapeutic modality that would best support a patient with a certain profile to achieve mental wellbeing
The exciting part of using big data analysis in mental health is for pre-diagnostic, or predictive assessments. With a large enough dataset containing the right data variables, health care providers could determine who may be more likely to experience mental health issues and disorders, and then based on this assessment, they could provide guidelines for therapeutic techniques, practices, and support to train the individual to deal with their mental and emotional challenges before their life is affected. People could therefore be trained to be more psychologically resilient before they experience a mental illness. In the case of suicide, the incredible benefits of such technology is life saving.
Psychological profiling could be based on the following data collection methods:
- Questionnaires and psychometric tests
- Emails, posts, tweets, blog articles
- Narrative-based assessments
- Voice patterns
- Patient health records
- Brain wave recordings
- Heart rate monitoring
- Breath rate monitoring
In building accurate psychological profiles, it is important to delve into the sometimes murky and messy world of thoughts, emotions, feelings, intentions, fears, hopes, aspirations, and attitudes. Internal states reveal more than what can be gathered through merely recording behaviours such as how many cigarettes or alcoholic drinks someone has per day, whether they take medication, or whether they exercise. Our behaviours and actions are driven by our internal state rather than the other way around, so these internal conditions are vital for effective profiling.
Suicide prediction, machine learning, and future wellbeing
Data analysis techniques, such as machine learning, have already been used in mental health for suicide prediction. Assistant Professor Colin Walsh and his team at the Vanderbilt University Medical Center have created machine-learning algorithms that predict the likelihood of patient suicide attempts. In a study of 5,167 adult patients with a claim code of self-injury, results were 86 percent accurate when predicting whether someone will attempt suicide within the next 720 days (almost two years), and 92 percent accurate in predicting whether someone will attempt suicide within the next seven days. Walsh’s groundbreaking work shows a tremendous capacity for data science to create positive outcomes for people suffering depression and anxiety.
The hidden potential of using big data in mental health is extraordinary. If we were able to harness psychological profiling derived from big data, assessments could be implemented in schools and the results could be used to support children as they grow into young adults. Children could, therefore, learn how to attain mental and emotional resilience from a very early age. Ultimately, this would lead to fewer adults experiencing mental illness and a greater level of individual and collective wellbeing. The future of our mental health is brimming with hope and optimism.