Category Archives: News

Big Data in Mental Health – Abstracts

flyer 140613

The speakers for Wednesday’s Big Data in Mental Health event have kindly made their abstracts available. Hope you’re all looking forward to it! Take a look below for details of the talks.

If you would like more information about the day, check out the event page or get in touch.


Michael Lynskey, King’s College London

Title: Integrating epidemiological, geographical and statewide administrative data

In Missouri, access to Sate-wide vital records (birth, marriage/ divorce, driver’s license and death records) has been used to enhance epidemiological research, primarily through identification of informative samples.  Now, given increasing capacity for linking to existing data sets and for geocoding locations of events we have the capacity to examine environmental influences on health outcomes at a population level. This talk will discuss the rationale for – and challenges in – integrating population wide data with geographic and epidemiologic data.


Maxim Ozipov, University of Oxford

Title: From questionnaires to objective data – continuous monitoring of mania and depression symptoms using mobile phone sensors

The diagnosis of mental disorders depends heavily on clinical
judgment. Numerous mental health scales have been suggested to bring
objectivity into the diagnostic process, however assessments require
periodic encounters with a patient and compliance with electronic
self-reports can be low. An automated assessment of mental health may
provide more timely interventions and decrease the severity of
significant events. Changes in sleep and activity patterns are
important symptoms of mood disorders and such changes can be
indicative of deteriorations in a patient’s clinical state. Using
smart phones we can continuously monitor changes in activity patterns,
as well as social interactions, thus potentially providing a low-cost
pervasive monitoring modality for the assessment of clinical status in
mental health patients. Here we describe an implementation of such a
system, as well as the first results and challenges arising from a
pilot study on healthy volunteers. We also propose a framework for
continuous symptom assessment in the presence of real-world
challenges.


Richard Jackson, King’s College London

Title: Concept extraction from clinical free text, and integration with research datasets.

In the course of clinical care, vast quantities of unstructured clinical notes are generated. These store and communicate vital pieces of information between stakeholders in patient care. In Mental Health, clinical notes are an especially important, due less reliance on discrete diagnostic categories relative to other health disciplines. However, additional uses of clinical data for the betterment of public health are not possible while it is in unstructured format. This talk explores the mass information extraction techniques the South London and Maudsley NHS Trust employs to derive meaningful, structured data from the Trust’s 20 million clinical documents, and the discoveries that result.


Amos Folarin, King’s College London

Title: Quantitative monitoring of the patient behavior and environment using small wearable devices and mobile phone sensors – a feasibility study on automated sleep measurement

Rapid and continued advancement in ‘omics technologies have provided us
with a wealth of high resolution information on the inner workings of
disease and genetic variability. A collateral rise in technological
advances and widespread availability of mobile phones has at the same
time been observed. These mobile technologies promise the potential of
gaining equally fine grained information on patient behaviour and
environment, especially when paired with additional small wearable
probes that report specific information on the user. The ability to
capture this data and report it back into the clinical record will be
of immediate benefit, such as providing clinical trials with high
resolution data for patient stratification, real-time response to
treatment, and directly improve patient care and efficiency of
intervention.



SLaM employs an electronic health record (EHR): the Patient Journey
System (PJS) and has implemented a de-identified derivative of this EHR
for research use:  the Case Register Interactive Search (CRIS).
Research projects can apply for access to CRIS data through a
well-developed governance framework. To give patients a more
interactive relationship with their clinical record, SLaM is using
MyHealthLocker, a system built on top of Microsoft HealthVault.
Clinicians can elect to share EHR data with their patients and patients
can submit data to their own health record and elect to share parts of
that record with their treatment team.



The Purple Robot application, developed by the Centre for Behavioural
Intervention Technology (CBITS) at Northwestern University, provides
both an interface to the on-board mobile phone probes (e.g. GPS, light,
temperature) and also a framework for integrating other mobile-aware
probes e.g. worn accelerometers, heart rate, perspiration, temperature,
oxygen saturation, blood pressure, blood glucose etc. Using this
platform, we have built a framework to enable patient reported outcome
(PROM) data from smartphone applications, smartphone sensors and other
sensor devices to be collected and analysed. Pertinent summary data can
then be exported into HealthVault and, through a MyHealthLocker widget,
patients can access and visualise their data and choose to share that
data with their care team or researchers.



In this talk I will discuss the use of this framework in the
development of a proof-of-concept application which tracks sleep
(quantity and quality).



Chris Hollis, MindTech, Nottingham

Title: NIHR MindTech HTC: Harnessing technological innovation to transform mental healthcare

MindTech (www.mindtech.org.uk) was established by NIHR in 2013 as one of eight NIHR Healthcare Technology Co-operatives (HTCs) in England. MindTech’s role is to drive forward technological innovation in mental healthcare by identifying clinical unmet needs and bringing together patients, clinicians, researchers and industry to develop and test a range of technological solutions to improve healthcare delivery and patient outcomes. User involvement in the design and testing of technology is central to our aims and we have established a national MindTech Patient Reference Group to support this work.

The growth and power of digital technology ranging from smart phone apps, sensors, automated facial, voice analysis and diagnostics is remarkable – and can provide enhanced self-management, remote on-line therapy, objective diagnostic and real-time monitoring for mental health conditions. Current research projects supported by MindTech include a randomized trial of computerised assessment of attention and activity in ADHD, automated facial and voice analysis to monitor mood, medication and self-monitoring apps linked to clinician decision support systems, investigating efficacy of remote on-line CBT and developing appropriate technology to live better with dementia.

The presentation will highlight both the opportunities as well as the challenges facing technological innovation and building the evidence-base for applications in mental healthcare.


Will Spooner, CTO Eagle Genomics

Title: The journey from 100,000 genomes to personalized medicine. Opportunities and challenges from an informatics perspective

In 2012 Eric Lander reflected: “we should remain unabashed about the ultimate impact of genomic medicine, which will be to transform the health of our children and our children’s children”. Although the molecular basis behind many diseases including several mental disorders has been uncovered, application of genome sequencing to their treatment is still very much in its infancy. Is that about to change? Genomics England Ltd (100% owned by the Department of Health) has been funded to sequence 100,000 Genomes. This project will leverage the unique position of the UK healthcare system and aims to provide an unparalleled resource for investigating clinical applications of genomics. Solving the informatics puzzle at the intended scale is challenging, and calls for specialised approaches; Eagle, a leading Cambridge-based bioinformatics services company, has been working on prototype software to tackle this problem that draws heavily on existing open source resources. The result is a robust, secure and scalable data processing and annotation platform with a distinct emphasis on metadata management that supports both primary (clinical) and secondary (research) uses of the resulting data sets.


Zina Ibrahim, King’s College London

Title: Multi-agent systems for randomized control trials in routine practice.

Results from randomized controlled trials (RCT) show that Methylphenidate is one of the  most effective drugs for the treatment of  Attention Deficit Hyperactivity Disorder (ADHD) with effect size ranging between 0.8 and 1.2. However, the results from RCTs do not fully represent long-term treatment or the heterogeneous population of patients with ADHD found in the clinical records. 

The problem is that in order to assess the long-term effectiveness of Methylphenidate, repeated measures of treatment outcome taken at the right time should be available for patients taking the drug. However, in real practice this is seldom the case. 

In order to improve the effectiveness of outcome measurements, and consequently the service to patients, we have devised a system for automatically request treatment outcomes from patients at regular intervals without clinician participation. 

The system is formed of a community of software agents which cooperate to collect treatment outcomes from patients at predefined intervals. The software operates by monitoring the Electronic Patient Journey System (EPJS) containing patient records for new entries of patients prescribed Methylphenidate. The system creates a personalised scheduled for each patient, prompting them to submit new assessments on regular intervals. The patients are able to submit the outcome through MyhealthLocker and are given in return a detailed Development and Well-Being Assessment (DAWBA) report based on their submitted outcome results.


Sean Maskey

Title: Big data, better information. A clinician’s wish list

Abstract: TBC


Maneesh Juneja

Title: How can Digital Health technologies improve outcomes, lower cost and improve care in Mental Health.

Digital Health tools & technologies promise to transform health & social care. Entrepreneurs around the globe are being encouraged to come up with new ideas to solve our biggest issues. The convergence of Big Data, Wearable Technology & the Internet of Things is creating not just new possibilities, but new challenges for healthcare systems and policy makers.

In this talk, Maneesh will go beyond the headlines, and share insights into how Digital Health is being used in Mental Health around the globe. His talk will also cover the implications of these new ideas, not just for providers and payers, but for patients and their families. He will also give a glimpse into the technologies heading our way over the next decade. 

At the heart of Digital Health is data, but who will own and control this new stream of patient data? How will it integrate into existing systems? What are the privacy & security risks associated with adopting these new tools? What data is of specific value in Mental Health? Why are Apple, Google and Samsung competing to develop global platforms for collecting, storing and sharing of health data? 


Symposium on Big Data in Mental Health

flyer 140613

We are excited to announce our upcoming symposium on Big Data in Mental Health.

The event will be held at ORTUS, Denmark Hill, on July 23rd and highlights will include Maneesh Juneja, Chris Hollis from MindTech, Will Spooner from Eagle Genomics and speakers from the University of Oxford, KCL and SLaM.

See the event page for more details: https://phidatalab.org/events/big-data-in-mental-health/

Registration on Eventbrite: https://www.eventbrite.co.uk/e/big-data-in-mental-health-biomedical-research-tickets-11667875931

PRQ

Algorithm design and strategy by Dr David Baker (Janssen). Implementation and application by Dr Steven Kiddle (KCL). For queries please contact steven.kiddle (at) kcl (dot) ac (dot) uk .

 

Pre-processing for Relative Quantification (PRQ) of TMT tagged LC-MS data is an R script written to pre-process mass spectrometry data. It performs the following steps:

Median normalisation to correct for labelling and MS-run variation (Step 1) is performed within each sample and gel fraction. This involves calculating the median of the ratios of all peptide intensities from one sample versus the corresponding intensities measured in the reference sample. All intensities relating to that sample and gel fraction are then divided by the median ratio. Ratio scores for each peptide are then calculated (Step 2) by calculating the ratios of the normalised data for each peptide by dividing it by the reference intensity. Ratios corresponding to the same source protein, peptide sequence and gel fraction are then summed. Protein level data is derived from these summed peptide scores (Step 3) by taking either the mean or median of all peptide scores from the same source protein and gel fraction. This protein level data is then collected across all sixplexs (Step 4).

A paper describing and applying PRQ has been submitted.

Open source script and readme available here

 

Post Doctoral Researcher in the application of Digital Technology

Post Doctoral Researcher in the application of Digital Technology

King’s College London

We are looking for an experienced Post Doctoral Researcher in the application of Digital Technology to Mental Health.

In the digital era we generate more data every 2 days than we did in the time up until 2003. This big data is being used by advertising, media, retail, finance and travel with medicine and healthcare lagging behind. The data held in electronic patient records and generated through modern digital devices are examples of the big biomedical data that we could use to develop more targeted treatment strategies. So, the challenge in e- and m- health is no longer data generation, the problem has shifted to data handling and the separation of signal from the noise to produce clear indicators of whether a clinical intervention is necessary. This is especially an issue in mental health as we work with less clear signals.

The exciting opportunity is now to use new technology to support more sophisticated models which might detect signals early enough to provide opportunities for effective interventions that keep people well.

This post will be based in our established BRC informatics group and work with the Patient and Carer Theme to develop shared informatics tools and infrastructure to enable data harmonization and analytics for use in discovering the personal signatures for needs for care using machine learning for example.

The role will contribute to the growing initiatives within the BRC clinical informatics, bioinformatics and biostatistics groups (https://phidatalab.org) whose others areas of focus include integrating genomics with electronic hospital patient records and cloud based patient owned records such as MyHealthLocker.

The post-holder will have an established track record in programming on the Linux OS, and will ideally also have quantitative research skills and/or previous experience of working with digital health data although this is not essential.

The post is funded for 18 months and the salary is grade 6 on King’s salary scale, currently £31,644 -£37,756 plus London allowance of £2323 per annum.

For an informal chat pleas contact Dr Richard Dobson, richard.j.dobson@kcl.ac.uk; Prof Til Wykes,til.wykes@kcl.ac.uk

To apply, please go to www.kcl.ac.uk/jobs.

Consent, Ethics and Data Security Workshop

We had a great turnout for last Thursday’s workshop looking at issues around using clinical data in research. I hope everyone who attended found it as useful as I did.

Someone asked for a copy of the worksheet from the morning session, so I’ve uploaded it here: Data Management Questions

Huge thanks to all of the speakers. I have uploaded slides to our slideshare account: http://www.slideshare.net/kclcompbio and linked to them below. I’ll upload summaries of the breakout sessions later this week.

If you have any questions or comments about the workshop, feel free to drop us an email or comment on this post.