FHS Research Project
Weekly Meeting 1
Friday 10/20/2017

1. Introductions

2. FHS Background

As the website for the Framingham Heart Study explains: 

Since our beginning in 1948, the Framingham Heart Study, under the direction of the National Heart, Lung and Blood Institute (NHLBI), formerly known as the National Heart Institute, has been committed to identifying the common factors or characteristics that contribute to cardiovascular disease (CVD). We have followed CVD development over a long period of time in three generations of participants. 

Our Study began in 1948 by recruiting an Original Cohort of 5,209 men and women between the ages of 30 and 62 from the town of Framingham, Massachusetts, who had not yet developed overt symptoms of cardiovascular disease or suffered a heart attack or stroke. Since that time the Study has added an Offspring Cohort in 1971, the Omni Cohort in 1994, a Third Generation Cohort in 2002, a New Offspring Spouse Cohort in 2003, and a Second Generation Omni Cohort in 2003. 

Over the years, careful monitoring of the Framingham Study population has led to the identification of major CVD risk factors, as well as valuable information on the effects of these factors such as blood pressure, blood triglyceride and cholesterol levels, age, gender, and psychosocial issues. Risk factors for other physiological conditions such as dementia have been and continue to be investigated. In addition, the relationships between physical traits and genetic patterns are being studied. 

In particular, the FHS began doing neuropsychological testing of its participants in 1999, and has been making digital recordings of the testing interviews since 2005. We have just begun a collaboration with Rhoda Au, the Director of Neuropsychology for the FHS, to explore the possibility that a more detailed analysis of these recordings might produce interesting results. Some researchers in Apple's Siri group are also involved in the project, as are Murray Grossman in Penn's Neurology department, Nadia Biassou from the NIH Clinical Center in Maryland, and Andrew Trister from Apple's Health Special Projects group.

According to the proposal that the FHS's executive committee has approved, the detailed aims of the collaboration are: 

Aim 1: Generate "gold standard" transcripts of 8000+ existing voice recordings as well as those being acquired through on-going neuropsy. testing of all FHS cohorts.
Aim 2: Analysis of the 8000+ existing voice recordings obtained between 2005-present using existing voice recognition and voice analysis software.
Aim 3: Build additional software to analyze the digital voice signals and generate novel cognitive metrics from latency and other behavioral characteristics.
Aim 4: For each neuropsychological tests as well as across tests, identify normative values for e-voice metrics stratified by age, education, sex, both individually and in combination (e.g., age x education; age x sex; age x education x sex)
Aim 5: Conduct factor and cluster analysis of e-cognitive metrics across neuropsychological tests to identify domain specific measures.
Aim 6: Determine e-voice metrics/profiles that differentiate between those with and without known AD risk factors, including but not limited to, ApoE, family history of dementia/AD, homocysteine, vascular risk factors (including metabolic), inflammatory markers.
Aim 7: Determine whether neuroimaging biomarkers are related to e-voice metrics/profiles. 
Aim 8: Determine whether incident change in neuroimaging biomarkers and neuropsychological tests are related to e-cognitive profiles.
Aim 9: Determine whether e-voice metrics/profiles can differentiate participants who are low to high risk for dementia/AD.
Aim 10: Conduct data driven analyses to identify e-voice metrics/profiles predictive of AD endophenotypes and risk for dementia/AD, in isolation and in combination with other health, lifestyle, biomarkers and genetic risk factors. 

The main task of the undergraduate research team will be to work on transcription of (a sample of) the recordings, using software and procedures that you'll be trained on. Depending on your interests and skills, there will opportunities to participate in other aspects of the project as well.

A background paper from the FHS NP team, along with copies of the current "Interviewer Test Sheet", "Test Administration Manual", and "Scoring Manual", can be found here.

3. Getting On Board

The basic hiring process: You...

  1. already had, or have applied for and received, a Social Security card
  2. have met with someone in Coordination at LDC to hand in hiring paperwork (and, if relevant, have provided verification of a work study award)
  3. have met with someone at LDC to complete an online I9 (if the I9 could not be completed at that time, you are still at this step)
  4. came to LDC to sign your offer letter, which describes conditions of employment
  5. have received an email from Coordination at LDC advising you how to log into your LDC email account and asking you to create a WebAnn account
  6. have received an email from Coordination with the subject line:  Hiring Process Complete (includes instructions on setting up direct deposit)

Then there are two steps of specialized training --

  1. CITI Human Subjects Certification (the "Social/Behavioral Research" track). This is an online course that typically takes a couple of hours. You'll get email with a link to the site.
  2. Training in the use of the transcription software and the transcription guidelines for this particular project. This will happen in person, in small groups.

4. The Data

See this file for a summary as of last July. New data is coming in at a rate of 50-100 interviews per month.

According to a recent email from Rhoda Au,

We have about 220 autopsy cases and we have identified those that have voice recordings.  A subset will also have antemortem MRI scans.  

Further, we have about 576 participants who are currently enrolled in our brain donation program, the majority of whom are older.  Some of these will have longitudinal recordings.  Keep in mind that these will also likely be people who have stored biological samples.  This means two things:  1) we likely have peripheral biomarker data already on them and 2) we have the possibility of building out a more comprehensive longitudinal biomarker database using stored samples.

5. Background Readings

We probably won't have time to discuss these, but you may find them useful in giving you a sense of the kinds of neuropsych research coming out of FHS.

Elias, Merrill F., Alexa Beiser, Philip A. Wolf, Rhoda Au, Roberta F. White, and Ralph B. D'agostino. "The preclinical phase of Alzheimer disease: a 22-year prospective study of the Framingham Cohort." Archives of neurology 57, no. 6 (2000): 808-813.

DeCarli, Charles, Joseph Massaro, Danielle Harvey, John Hald, Mats Tullberg, Rhoda Au, Alexa Beiser, Ralph D’Agostino, and Philip A. Wolf. "Measures of brain morphology and infarction in the framingham heart study: establishing what is normal." Neurobiology of aging 26, no. 4 (2005): 491-510.

Au, Rhoda, Joseph M. Massaro, Philip A. Wolf, Megan E. Young, Alexa Beiser, Sudha Seshadri, Ralph B. D’Agostino, and Charles DeCarli. "Association of white matter hyperintensity volume with decreased cognitive functioning: the Framingham Heart Study." Archives of neurology 63, no. 2 (2006): 246-250.

Bondi, Mark W., Emily C. Edmonds, Amy J. Jak, Lindsay R. Clark, Lisa Delano-Wood, Carrie R. McDonald, Daniel A. Nation et al. "Neuropsychological criteria for mild cognitive impairment improves diagnostic precision, biomarker associations, and progression rates." Journal of Alzheimer's Disease 42, no. 1 (2014): 275-289.

Jun, Gyungah, Carla A. Ibrahim-Verbaas, Maria Vronskaya, Jean-Charles Lambert, Jaeyoon Chung, Adam C. Naj, Brian W. Kunkle et al. "A novel Alzheimer disease locus located near the gene encoding tau protein." Molecular psychiatry 21, no. 1 (2016): 108.