Just  real  quick,  I'm  going  to  give  the  usual  reminder. Obviously,  I  am  neither  Keith  nor  Richard. We've  really  got  the  B  team  out  today. Both  of  them  had  personal  things  to  attend  to, but  everyone  should  remember. Please  do  make  sure  to  clean  up  after  yourselves  and  make sure  you  silence  anything  that's  going  to  make  noises. I  already  managed  to  catch  my  laptop  making  noise, so  it  happens  to  all  of  us. Today's  talk  is  going  to  be  really  fun  and  exciting. I'm  sure  I  don't  have  to  introduce  Rosa  to  anyone  here, but  she  leads  the  Ubc  Health  and  Wellness  Lab currently  and  does  fascinating  research. And  so  you're  here  to  listen  to  her  talk,  not  me. So  I  will  let  her  go  ahead  and  get  started. Yeah.  So  thank  you  so  much. You  know,  there  were  a  couple  of  e  mails  that  came  in. You  know,  this  and  that.  Rose.  I'm  like,  It's  okay. I  require  no  introduction.  There  you  go. I'll  make  myself  at  home  Today, I'm  going  to  talk  to  you  about  forms  of accountability  at  theater  section  of  science  and design  implications  from  ecologies of  care  studies  and  PTSD  and  diabetes. Um,  first  of  all, I'm  going  to  give  you  a  general  introduction  to my  research  and  in  the  area  of  health  and  wellness. I'm  going  to  argue  that  computing  systems can  address  fundamental  problems in  chronic  care  management. I  teach  HTC  one  this semester  and  I've  decided  that  I'm  optimistic, despite  everything  else  that all  of  the  critical  theories  tell me  I  should  be  thinking  about. But  really,  this  presentation is  an  opportunity  to  discuss the  role  that  accountability  plays  in  conducting research  at  the  intersection  of  science  and  design. I'm  going  to  talk  to  you  about  some  of the  theories  that  drive  my  accountability. Then  I'm  going  to  talk  about  lessons  I've  learned  over these  past  15  years  in  HCI  and  future  work. But  first,  my  title  is  pretty  convoluted. Can  I  see  a  show  of  hands  for  people  that understood  why  my  presentation  title  is  so  convoluted? At  least  one  person  that  took  HTC  one, maybe  my  title  is  really  homage  to  Bill  Gaver, who  wrote  a  chapter  in this  really  great  book  that's called  Ways  of  Knowing  an  HCI. In  there  he  talks  about  science  and  design  and the  implications  of  different  forms  of  accountability. And  he  defines  accountability as  expectations  of  what  must  be  defended. And  the  way  that  these  narratives  in  science  and design  are  form  and  how  they  drive  our  endeavor. In  this  chapter,  he  characterizes  what  makes  science and  design  different  based on  the  activities  that  are  essential  for  each. Now  he  also  says,  look, we  could  talk  about  all  the  things  that  are the  same  about  science  and  design,  right? They're  both  empirical  disciplines. Or  rather,  if  we  think  about  HCI and  computer  science  as  examples. But  he  really  argues that  when  we  focus  on  the  similarity, it  really  obscures  the  very  features  that  give each  endeavor  its  specificity  and  potency. He  emphasizes  that  science  and  design each  has  its  own  logic,  motivation,  and  values. And  that  if  we  were  going  to  put these  kind  of  families  of  activities  together, then  we'd  say  that  science  uncovers  what  exists, and  design  creates  the  new. And  his  chapter  is  really  nice. I  urge  you  to  read  it  and  maybe we'll  have  a  conversation  afterwards. The  first  half  is  basically drawing  these  differences  between  science  and  design. The  second  half  is  him explaining  what  his  type  of  design  is. And  I  will  cover  that  much  more  lightly, but  it  really  this  understanding  that what  it  is  that  we  do  really drive  against  that  accountability. I  discussed  a  little  more. After  all,  I  guess  this  would be  the  positionality  part  of  the  talk, where  I  tell  you  that,  you  know, why  am  I  so  fascinated  with  this  chapter? And  so,  I  don't  know  if  you'll  be  able  to  see  it, but  the  idea  was that  Gabor  and  I  share  an  origin  story,  right? So,  in  this  chapter  he goes  on  to  say,  and  it's  an  e  mail, I  actually  had  the  courage  to  write  him  and  say  that, that  I  was  going  to  give  this  talk and  really  talk  about  his  paper. But  really,  you  know, we  shared  the  same  origin  story. He  was  a  cognitive  neuroscientists that  moved  into  design. I  was  a  cognitive  neuroscientist  that  moved  into  design. I  thought  about,  you  know, these  implications  for  the  kind  of  work  that  we  do. And  he  was  very  nice  and  I  made  his  day. And  once  again,  it's always  good  to  put yourself  out  there  because  if  nothing  else, you  can  add  it  to  your  talk.  All  right? So  my  training,  as  I  mentioned, in  experimental  psychology,  right? We  study  human  thought,  perception  and  behavior. Of  course,  science  is  a  method  of  choice  where  we make  observations,  we  make  predictions, we  test  these  predictions, we  develop  interventions  to  improve  the  human  conditions. Then  we  use  these  theories  to  drive  our  research. And  I'm  going  to  define  theories  as  a  set  of  principles that  organize  our  observations and  also  propose  predictions. And  so,  you  know,  Gabor  would  say,  well, of  course,  rosa  science and  the  way  that  you've  just  described, it  has  this  epidemiological  accountability  science is  really  concerned  about how  knowledge  is  established,  right? How  do  you  know  what  you  claim  to  know? It's  really  concerned  and  driven  by  this  kind  of  process, how  rigorous  the  process  is, how  rational  the  process  is. Really,  if  you  compare  it  to  design, this  notion  of  something  being  interesting  or impactful  is  not  necessarily  definitional. Gabor  goes  on  to  say  that  progress  in  science is  either  inductive  or  hypothesis  driven. And  of  course,  when  you  think  of  science, you  think  it  has  to  be  replicable. There  has  to  be  objective, it  has  to  be  generalizable. You  have  to  move  towards  establishing  or expanding  on  scientific  theory. But  then  I  moved  into  HCI  research  here,  right? This  discipline  studies  how people  interact  with  computers. But  at  some  level  we're  really designing  systems  that  are  useful, usable,  and  enjoyable, that  allow  others  to  meet  their  goal. The  idea  is  a  human  centered  computing. We  take  it  a  little  further  to  say, well  actually  it's  not  just about  the  human  and  the  machine, it  really  is  about  how these  computational  artifacts  affect and  transform  the  human  experience  and  culture. And  that  loop  of  us  also  transforming  the  technology. If  I  think  about  the  method  that  I  use  most, I'm  going  to  talk  about  a  user centered  design  approach  where I'm  trying  to  understand the  context  by  doing  needs  assessment, design  ideation,  prototyping,  evaluation. I'm  talking  to  people  that know  all  of  this  so  you  don't  have  to take  my  course  courts, but  you  can  if  you  want  in  HCI. Our  contributions  are  pragmatic  and  futuristic,  right? We  come  up  with  novel  interactions, with  novel  interfaces,  with  novel  systems, but  also  we  generate  ideas. We  start  to  talk  about  what are  those  design  implications, What  are  those  design  guidelines? And  here  again,  Gabor  would  say, where  we  just  talked  about  the  kind  of epistemological  accountability that's  inherent  in  science. When  we  have  design,  we  have this  aesthetic  accountability,  right? It's  going  to  have  to  be based  this  question  of  does  it  work? Central  design  is  really  based  on  the  problem  solving production  value  and  the  ability  to integrate  function  form,  material  culture  motion. Again,  entire  socio  technical  experience. And  we're  going  to  say  that  design  is  successful,  right? When  it  validates  new  methods, it  drives  conceptual  perspective. But  again,  the  methodology  is  not lockstep  the  way  it  is  in  science, and  it's  not  critical whether  for  how  it's  going  to  make  or  break  the  design. Then  Gabor  goes  on  to  say  that  progress  is driven  by  the  ability  to  meet  the  context,  right? So,  does  it  work  by  the  ability  to  create and  constrain  these  novel  design  spaces? But  whereas  in  science  we  talked  about  generalizability, replicability,  objectivity,  in  design, we're  going  to  have  different  values. We're  going  to  be  driven  by  how  it  meets  the  individual, how  it  resonates  with  the  context, how  evocative  it  is, how  illuminating  it  is. Where  does  science  and  design  meet  me  in the  health  and  wellness  research  I do  and  the  artifacts  I  build. Because  for  me,  the  artifacts  I build  have  to  be  useful  and  usable. But  they  also  have  to  be  general, replicable  and  objectively  efficacious. If  I'm  going  to  build  something for  somebody  that  has  asthma, diabetes,  or  PTSD,  at  the  end  of  the  day, I  want  to  know  that  it's  actually  going  to  improve this  condition  for  me. It  means  that  I  can't  just  leave  it  at  the  design, but  I'm  going  to  have  to  deploy  it. We  talk  about  intersectionality and  why  it's  critical  in  health  research. Currently  in  the  US,  health  spending  is about  70%  of  the  gross  domestic  product. It's  going  to  be  about  a  fifth. It's  a  huge  problem. Of  course,  it's  not  just  about  the  money. About  7.10  deaths  in  the  US  are attributable  to  chronic  illness. What  does  it  mean  that  I  do  this  work  at the  intersection  of  science  of  design? Well,  it  means  that  I  build computational  systems  because  they provide  a  way  to  scale human  effort  over  time  and  space,  right? If  I  talk  to  a  psychologist  or a  pomonologistorpodiatrist, they  can  meet  one  person  at  a  time. But  really  with  the  systems  I  build, we  can  reach  many  people. Again,  the  other  thing,  the  other  place  where I  bring  my  science  to  bear is  really  in  using the  psychological  theories  and concepts  to  conduct  HCI  research. Again,  when,  if  the  premises  that  humans, her  computing  can  improve  the  human  condition. Then  for  me  at  some  point  I  have  to have  evidence  from  both  science  and  design. Again,  it  must  be,  it must  be  useful.  It  must  be  engaging. And  I  have  to  be  able  to  show  you,  right, that  I  can  improve  health  and  wellness  outcomes. So  what  I'm  going  to  do  in  the  rest of  the  talk  is  really  give you  kind  of  three  examples of  how  I've  done  this  in  my  work. I  will  give  you  the,  the  most  mature research  which  is  the  work that  I  did  in  pediatric  asthma. And  by  the  way,  that's  right  when  I  was moving  from  science  to  design. And  so  of  course, I  was  going  to  do  a  random  control  trial. I'm  going  to  tell  you  about  the work  that  I'm  doing  now  and post  traumatic  stress  disorder. Tell  you  a  little  bit  about  how  that design  really  informs  what  I  do. Then  I'm  going  to  tell  you  about the  newest  work  that  I'm  doing in  diabetes,  All  right? I've  told  you  that  chronic  conditions are  such  a  big  problem. If  we  think  about  what these  chronic  conditions  have  in  mind, we're  going  to  have  these  major  gaps. First  is  engagement. How  is  it  that  we  can  get  the  user to  act  in  a  way  that's  going  to  improve  their  health? How  is  it  that  we  can  use  the  current  treatments  and practices  and  really  build technology  in  a  way  that  supports  the  user? In  chronic  conditions  like  diabetes,  like  asthma, people  see  their  Dr.  three  or  four  times  a  year, but  there's  never  any  data about  what  happened  between  those  visits. You've  all  had  the  experience. You  go  to  the  Dr.  and  the  Dr.  says,  how  are  you? And  you  say  fine,  even  though  you  had this  killer  migraine  three  weeks  ago, you  just  can't  remember. Again,  one  of  the  things  that  I  want  to do  with  my  technology  is  I  want  to understand  what  happens  between  a  visit  for  the  provider. Where  the  provider  can  be,  again, it  can  be  a  physician,  the  therapist,  a  podiatrist. Then  once  you're  at  the  point  of  care, how  do  we  mediate  communication? What  would  the  provider  want  to  know? What  would  you  want  to  tell  the  provider? How  do  we  drive  self  advocacy  through  our  system? Again,  over  the  past  50  years, I've  really  designed  systems  to support  a  variety  of  stakeholders. Not  just  the  individuals that  might  have  a  given  condition, but  their  families  and  professionals  that  care  for  them. Again,  I  use  ubiquitous  interfaces  because  that's how  we  know  that  they  can actually  work  in  the  real  world. Central  to  all  of  this  is  that  collaboration with  domain  experts  are  key,  right? That  is,  I  need  to  understand what  the  gold  standard  of  care  is,  what's  the  science. Then  I  need  to  understand  on  the  design  part, how  will  I  make  sure  that  what  I  build  works? I've  had  a  great  time  here  in  Atlanta. I've  been  able  to  work  with Children's  Health  Care  of  Atlanta,  for  example, looking  at  AR  education  games  in  the  emergency  room. I've  worked  a  lot  with  Emory  Hospital, They  continue  to  be  really great  colleagues  and  now  friends. After  these  15  years, I've  also  had  the  great  pleasure  of  working  with  the  CDC. One  of  our  crews  in  our  lab  is  that  we designed  the  first  CBC  milestone  tracking  app. For  the  first  version,  there  were  over 700,000  downloads  in  the  first  three  or  four  years. But  again,  I  tell  you  about  my  collaborations and  I  tell  you  about  the  methods  I  use. But  really  at  the  intersection of  science  and  design  for  me  is moving  beyond  those  traditional  user  centered design  approaches  to  using and  leveraging  different  theories. And  I'm  going  to  talk  to  you  today  about the  health  belief  model and  also  the  ecological  systems  theory. We  talk  about  a  chronic  disorder. How  are  we  going  to  operationalize  it? Well,  it  has  to  last  more  than  one  year, and  it  has  to  lead  to  limitations  in  function, activity,  or  social  role. There's  some  dependency  on  medicine, the  required  education,  beyond  what's  expected  of  an  age. This  is  exactly  what  happens  in  pediatric  asthma. Again,  looking  at  that  arc  from  going  all  the  way  from science  and  design  in  a  mature  manner, I'm  going  to  tell  you  about  that  work. As  you  might  know,  asthma is  characterized  by  inflammation  of the  airways  that  narrows them  and  makes  it  hard  to  breathe. It's  the  most  common  pediatric  disorder  in  the  world. If  I  want  to  know  how  do we  improve  health  outcomes  in  children  with  asthma, what  technology  might  we  use? Well,  I  think  even  now  text  messaging  was the  most  popular  way  to  that  individuals  in 13  to  17  age  range  communicated. If  I  was  going  to  have  to  pick  a  method  to say  that  I  can  improve  a  health  outcome, right?  You  know,  what  would  I  do? How  would  I  build  it  here? We  use  theory  to  drive  the  intervention  features. And  then  if  I  wanted  to  be  able  to make  that  scientific  claim  about a  causal  relationship  between  the  system I  deployed  and  children's  health  outcomes, then  I'd  have  to  use  a  random  control  tile. And  so  this  is  the  work that  I  did  with  using  the  health  belief  model, and  it  really  is  what  variables can  be  used  in  behavior  change  intervention. And  here  we  have,  again, back  to  my  bias  trained  as  a  cognitive  neuroscientists. This  idea  that  these  beliefs  are central  in  the  way  that  we  act  in  the  world. The  idea  was  that  we  were  going  to  take  some  of these  modifying  factors  like  knowledge  about  asthma, we  were  going  to  t  test this  hypothesis  about  symptoms,  right? The  idea  is  that  if  you  don't  think  you're  sick  enough, why  would  you  take  your  meds? And  so  the  idea  was  that  we  were  going  to  try to  send  out  these  messages that  would  get  people  to  think  about  their  condition. Can  text  based  intervention, using  these  health  belief  model features  improve  health  outcomes? Right?  So  is  it  useful?  Can  we  prove  it? Will  patients  engage  with  the  system  between medical  appointments?  Right?  Is  it  usable? Basically,  we  had  a  very  like  the  little  system that  could  kids  would  either  get  a  symptom  question. In  the  past  four  weeks,  did  you  wake  up at  night  wheezing  and  coughing, or  they  get  both  a  symptom question  and  a  knowledge  question. The  knowledge  question  would  just send  out  a  little  query. Everyone's  asthma  is  the  same,  true  or  false. And  then  depending  on  how  the  kid  responded, we  would  give  them  a  little  factoid  about  asthma. This  was  done  in  a  randomized  control  protocol over  a  three  month  period. Again,  in  children  have  moderate  severe  asthma, either  Dr.  three  or  four  times  a  year. And  that's  when  we  met  them.  Then  we  wanted  to, of  course,  help  the  physician  at  the  point  of  care. And  so  we  visualize  the  text  responses. The  doctors  had  access  to  the  dashboard,  right? It  allowed  for  data  driven follow  up  in  between  care,  so  they'd  get  pinged. If  the  kids  response showed  that  they  had  gone  over  to  the  red  side, that  is  that  they  might  be  at risk  for  an  asthma  exacerbation. We  also  had  all  of  the  responses  related  to their  asthma  knowledge  and  symptoms  and the  idea  that  this  would  prompt that  point  of  care  interaction. When  we  did  this  study  where  we looked  at  children  that  were  13  to  17, we  found  that  the  symptom  and  knowledge  that if  a  kid  got  a  text  message  a  day  right, they  actually  had statistically  significant  improvement  in  lung  function. All  they  were  doing  was  responding  to  text,  and  yet  this, their  improvement  in  lung  function, it  caused  their  improvement  in  quality  of  life. One  of  the  things  that  we  found  was that  this  was  the  case  when  we worked  with  children  that  had  private  insurance. Most  HDI  researchers  would  have  said,  great  it  works. But  again,  those  scientific  battling  for me  came  back  in  the  way  of  understanding, Can  we  replicate  result? Can  we  extend  the  result? For  example,  we  had  done  a  symptom  only  condition. But  what  if  we  did  a  knowledge  only  condition? But  also  what  are  those  conceptual  factors? Will  the  system  actually work  with  different  demographic  groups? Once  again,  that  premise  that  we can  improve  the  human  condition, that  good  science  can  drive  health  equity. For  me,  it  was  understanding, does  it  generalize  to  other  demographic  contexts. We  know  that  asthma  disproportionately affects  poor  individuals. We  know  that  black  children  are  five  times more  likely  to  die  from asthma  than  their  white  counterparts. And  so  we  did  the  series  of  studies  and children  that  were  low  SES  is  they  had  private, pardon  me,  they  had  public  insurance. And  so  we  knew  that  they  lived  in  a  household that  had  limited  funds. We  found  that  we  were  able  to  replicate  the  results. Is  that  when  in  my  favorite  title  for  a  paper, that  in  fact  the  text  message  today  does  keep the  pulmonologists  away  for children  that  have  public  insurance. Importantly,  we  that  when  we  had  these  children. Um,  that  had  public  insurance, they  would  have  improved  lung  function, even  when  we  send  a  text  message  every  other  day. Again,  back  to  the  scientific  accountability, the  patient's  lung  function  improved in  a  statistically  significant  manner. So  the  system  was  useful,  right? This  brings  up  Gabe's  notion  of  definiteness, that  we  know  that  something  is  precise  and  quantifiable. Then  there  was  also  the  design  accountability. The  system  was  useful. Patients  were  willing  to give  us  information  over  an  extended  period  of  time. All  right,  now  I'm  going  to  talk  to  you  about my  ongoing  research  in  post  traumatic  stress  disorder. Here,  this  is  much  less  mature  work. We're  in  what  Gabriel  might call  the  productive  indiscipline, right,  where  we're  not  bound  to  doing these  lockstep  experimental  studies that  are  going  to  give  us  some  definitive  answer. Here  I'm  going  to  talk  to  you  about the  ecological  systems  theory. This  is  the  developmental  psychological  theory. I've  used  it  to  think  about  how  can  technology be  situated  in  an  individual's  life. The  ecological  system  theory  was  proposed  by Uri  Bronfenbrenner  and  he  defined  development,  right, change  through  time  and how  the  individual  interacts  with  people  in different  settings  and  the  degree  to which  the  individual  can  alter  those  interactions. And  one  of  the  first  papers  I  wrote when  I  came  to  HCI  was  really  thinking about  how  can  we  go  from  this  typical  social  science, complicated  understanding  how  can we  flatten  it  so  that  can  be used  to  understand  the  role  of technology  design  in  children  that  had  asthma. This  was  my  students  early  on. What  I  say  is  that  the  power  in  using this  theory  is  that  it  really  defines the  setting  that  envelops  the  individual,  right? It  reminds  the  designer  about the  user's  entire  ecosystem. Again,  the  set  of  principles  that  are  going  to really  put  us  in  a  lockstep  manner to  understand  the  context. And  then  that  this  ecological  approach  can help  us  understand  where  to  target  technology. Over  the  course  of  those  ten  years, we  did  a  lot  of  studies  that  we're really  trying  to  figure  out. Not  technology  for  children  with  asthma, but  those  that  were  for the  child  and  the  family  in  the  community. Again,  one  of  the  things  about these  theories  is  that  they  are  generalizable. And  we  were  able  to  use this  ecological  systems  approach  to study  adults  with  post  traumatic  stress  disorder. Ptsd  is  a  trauma  and  stressor  related  disorder characterized  by  trauma  re, experiencing  by  avoidance  of  trauma  related  situations, by  negative  alteration  in  mood  and  negative  arousal. We  focused  on  veterans  because  about  20%  of  veterans of  the  Iraq  and  Afghanistan  conflict meet  diagnostic  criteria  for. And  then  you  know  that  I'm  in  design because  I  have  a  quote,  right? I'm  left  with  basically  nothing  to  trap  in  war, to  be  at  peace,  to  damage,  to  be  at  war. Once  again,  I  want  to  start  from the  premise  of  what  is  the  gold  standard  of  care. I'm  not  going  to  say,  what can  I  build  in  this  amorphous  space, but  I'm  really  going  to  constrain  it. The  gold  standard  of  care  for PTSD  treatment  is  prolonged  exposure  therapy. Right  here,  clinicians  teach  individuals  to gradually  approach  their  trauma  related  memories, feelings,  and  situations. One  of  the  things  that  they  do  during therapy  is  they  train  veterans, or  they  train  patients  really  on  how to  quantify  their  stress. This  subjected  unit  of  distress  is  really  important. They  train  these  individuals  on understanding  how  their  stress  level  is  changing. One  might  be  no  stress  and  100  is  inca,  incapacitating. What  happens  during  a  PE  therapy? Well,  there's  a  clinical  session, you  have  the  traditional  talk  therapy, where  the  patient  is  going  to voice  the  narrative  of  the  traumatic  event. Right?  They're  going  to  voice this  imaginal  narrative  again, because  what  they're  trying  to  do  is  get  the  patients  to understand  that  their  fears  are  unfounded. They're  going  to  ask  them  to  do  homework. One  homework  is  basically  listen  to  yourself  narrating This  imaginal  narrative. The  other  one  is  actually  confronting these  stress  provoking  stimuli  in  the  real  world. They  call  these  in  vivo. I'm  going  to  show  you  an  example of  an  imaginal  narrative. This  is  a  mock  narrative, it's  not  one  that  is  real, but  rather  that  curated  from  many  conditions. We've  heard  in  this  particular  narrative, something  very  terrible  happens  to,  to  a  person. They  basically  are  trying  to  escape  from  attack. They  end  up  harming  a  lot  of  children. And  again, this  feeling  that  they  could  have  done  something. But  the  main  idea  that  I  want  you  to  get  from this  is  that  you  have  the  narrative. And  then  there's  the  homework. So  the  homework  would  be,  go  home  and  listen  to  this  for a  half  an  hour  or  until your  subjective  distress  dropped  to  about  half. And  then  the  other  part  is  going  to  be  every  day,  right, on  Monday,  you're  going  to  go  to sit  in  the  park  with  a  friend. You're  going  to  be  there  for  30  min or  until  your  stress  decreases. Or  you're  going  to  go  to  the  park  by yourself  or  you're  going  to  sit  at  the  park  and eventually  you're  going  to  be  at  the  park  on Saturday  afternoon  when  all  the  kids  are  out  there. The  good  news  is  that  it's the  gold  standard  of  care  and  recovery  is  very  high. The  bad  news  is  that  it's  very hard  to  actually  deliver  it. Some  of  the  trouble  is  that  it  relies on  patient  self  support.  Right? So  far,  there  had  been  this  inability  to  collect objective  information  during  and between  those  clinical  visits. It  relies  on  clinical  intuition. Did  they  go  to  the  park,  stay there  for  30  min?  Were  they  engaged? There's  limited  decision  support  tools that  promote  guidance  of  proper  therapy, practice  and  reflection  towards  recovery. Then  that  because  we  don't  have  this  information, there  aren't  ways  to  understand  how to  improve  the  therapy  and  improve  the  delivery. We  were  very  excited  to  get an  NSF  award  in  Smart  and  Connected  health, where  we  were  trying  to  build  a  computational  solution  to improve  the  delivery  of  PE  and the  patient  practice  during  PE  therapy. The  idea  that  we  were  going  to  innovate  here, we  were  going  to  build  a  ubiquitous  system  to collect  the  data  during  the  PE  therapy. We  were  going  to  build  these human  centered  interfaces  that we're  going  to  allow  it  to  be useful  and  usable  to  both  clinicians  and  patients. And  then  that  Thomas  was going  to  be  able  to  work  his  magic  and  it  was going  to  allow  us  to  predict  how these  clinical  factors  might  affect  individual  recovery. Here,  right,  we  have  this  classic  opportunity  for  design. Ga  says  there's  this fashioning  something  new  that  works  in  the  world. And  here  we  can  see  that  the  theory, right,  so  the  science  is  really  driving the  design  that  we  work  with  here. We're  using  the  ecological  systems  theory to  both  create  and  constrain  new  artifacts. So  there  are  these  proposals  of  what  we  might  build. These  proposals,  according  to  Gabor, are  going  to  lead  to  progressively more  focused  design  explorations. There  really  are  these  conversations  with  materials. What  methods  are  we  going  to  use? Once  again,  we're  free  from  having  to  do these  lockstep  scientific  experiments. In  fact,  since  we  got  the  award, we've  been  able  to  study  veterans  ecologies  of  care. Trying  to  understand  what  are  some  of  those  barriers  that veterans  experience.  What  do  they  think? What  do  their  civilian  counterpart and  families  and  friends  think  would  do  the  clinicians? Then  from  there,  we  can  talk  about  what  are the  design  implications  of  these  results? What  tools  do  they  need? Right?  We  found  that  they  wanted  to  be  able  to provide  feedback  about  engagement  with  clinicians. They  wanted  to  be  able  to  collect data  from  outside  their  clinical  session. And  they  wanted  to  build  a  way  of  understanding  how others  might  be  able  to  inform  the  therapy. We  were  able  to  build  and  try  to  understand  what  we had  initially  called  this kind  of  social  sensing  or  social  scape. And  through  another  series of  studies  that  had  different  phases, again  a  very  different  methodology. Clinicians  trusted  others, and  veterans  were  presented  with the  storyboard  about  what a  social  sensing  or  an  informant  system  might  look  like. From  there,  we  were  able  to  get these  design  guidelines  that  emphasize  the  familiar. People  want  to  emphasize  the familiar  in  the  positive  when  they  give  feedback. But  when  they  receive  feedback, veterans  want  to  control  who actually  gets  to  provide  the  feedback. But  there's  also  the  need  to understand  how  it's  providing  the  feedback  actually. Affecting  these  relationships. And  then  there's  the  logistics. How  do  we  customize  the  cadence  of the  feedback  that  the  trusted  others  are  giving? When  we  think  about  the  kind  of  work  that  we  do  and the  selection  bias  that  we  might  have  when  we  have people  that  are  willing  to  be  in  our  studies,  right? This  really  drives  different  sets  of  questions. Working  with  GOA,  we  were  able  to  look  at, trying  to  get  a  broader  understanding of  what's  going  on,  really. We  looked  at  the  PTSD  or  rather  the  veteran  sub  red, to  try  to  understand  what  is it  that  veterans  actually  want. We  ask  them  about  PE. But  if  we  actually  analyze  the  veteran  sub  red, they  actually  care  more  about  health  finance, family,  and  navigating  the  governmental  support. Again,  how  do  we  provide the  right  structure  for  veterans to  get  the  care  they  need? But  again,  eventually,  right  for  my  type  of  research, I  want  to  be  able  to quantify  precisely  what  it  is  that  we  gain  by, by  building  the  system. And  the  third  aim  of  our  proposal  is  to  do  just  that. To  understand  the  various  data  types  we  collect. Right?  Eventually  what  the  machine  will  do  is  we  take the  subjected  unit  of  distress  and  what we  try  to  understand  is  eventually, can  we  predict  the  units  of  distress? Then  how  do  we  give  those  machines  that feedback  to  be  able  to  understand, perhaps  not  just  the  individual,  um, recovery,  but  how  the  program  is working  in  general  or rather  within  this  particular  segment  of  veterans. But  also  how  do  we  improve  these  care  models? One  of  the  things  we  found  in  the  colleges  of paper  is  one  of  the  common  reasons people  drop  out  is  because  our  veterans  drop  out is  because  they  find  that  they  don't  have a  good  match  with  their  clinicians. Then  the  question  is,  can  we  do  that? We  have  our  system, our  interfaces  and  we  hope  to deploy  them  this  next  semester. All  right. I've  told  you  about  the  arc  of  being  able  to do  random  control  trials  and  the  the  pediatric  betting. I've  told  you  about  how  my  work  progress  within PTFE  in  the  last  four  years  of  research, and  I  want  to  tell  you  a  little  bit about  what  I'm  doing  now. I'm  working  in  the  area of  diabetes  and  in  particular,  fouls. Once  again,  this  is  an  incredibly  important  area. As  you  might  know,  about  10%  of adults  are  diagnosed  with  diabetes  and 1.3  are  pre,  diabetic,  diabetic  fools. This  terrible  complication  that we  can't  as  scientists  get  our  arms  around, for  example,  the  mortality  rate after  a  person  has  an  amputation  is about  70%  Within  five  years  of  having  an  amputation, we  have  people  perish. Of  course,  this  is  in  underserved  communities, both  African  American  and  Hispanic. And  we  started  with  the  numbers  just  for  context, this  is  basically  $100,000,000,000  problem. About  a  third  of  diabetes  expenditure is  in  diabetic  foot  Al. All  right,  then  what  am  I  going  to  do? Well,  our  goal  in  the  next  year  is to  basically  understand  the  goal  ten  gold  standard of  care  for  diabetic  foot  Al. This  is  a  full  foot  exam. Traditionally,  podiatrists  will  measure  the  skin, they'll  measure  the  blood  circulation, they'll  measure  sensation,  and  they'll measure  the  soundness  of  the  musculoskeletal  frame. And  we  are  in  fact  building  a  app  to  do  that. I  want  to  thank  we  were  lucky  enough  to  get the  Emery  I  humanity  grant and  we've  seen  ten  of the  200  people  that  we  were  going  to  get  data  from. All  right,  well  what  does the  diabetic  ulcer  computational  system  look  like? Well,  no  surprise. Right,  So  far  we're  really  thinking  about how  do  we  build  a  system  for  the  patient. How  do  we  bring together  what  we're  calling  foot  soldiers?  Right. People  that  will  help  them  collect the  data  or  the  video  data. We're  thinking  about  what  are the  algorithmic  implications. Most  of  the  work  in photography  and  images  is  on  white  skin,  right? One  of  the  things  we're  trying  to  build is  the  biggest  database of  different  skin  colors. Then,  because  we've  learned  so  much  about being  able  to  move  beyond the  people  that  come  to  our  studies. How  can  we  learn  about  people's  practices in  Read  It  and  Twitter  again? Right.  We're  going  to  be  able  to leverage  what  we  got  from Pe  to  be  able  to  collect  the  data  for  this  study. So  just  to  recap, I've  given  you  examples  of my  research  at  the  intersection  of  science  and  design. I've  told  you  about  the  work  that  I've  done  in  asthma, and  I've  been  able  to  prove  that  there's a  causal  relationship  between  children  getting text  messages  and  their  improvement  in  lung  function. I've  told  you  a  little  bit  about the  prolonged  exposure,  continuing  sensing  system, and  I  hope  to show  that  it  actually  leads  to  more  efficacious  therapy. I've  introduced  you  to  the  newest  research  area, which  is  a  diabetic  foot  alters.  Stay  tuned. I'm  going  to  just  take  a  couple  of  minutes  to  reflect  on what  I've  learned  from  using theory  to  influence  technology  design. So  that  intersection  of  science  and  design  first, when  you  use  the  theory,  right, I  say  that  the  resulting  systems are  inherently  generalizable. I  told  you  about  the  work  we  did  in  asthma, but  we  were  also  able  to  use the  text  messaging  system  for  children  with  diabetes. We  were  able  to  use the  same  system  looking  at  congenital  heart  disease, again,  in  this  iterative  framework.  We've  learned  a  lot. We  learned  that  when  it's  children facing  some  negative  outcome  years  from  now, the  awareness  is  important  and it'll  get  them  to  drive  behavior. However,  when  we  were  looking  at  adults, these  are  people  that  are  40 that  are  faced  with  these  symptoms. It  actually  made  them  more  anxious  and not  any  more  likely  to  take  their  me  Again, these  studies  are  really  important  to understanding  the  context  of  when  these  systems  work. I've  told  you  and  I've  shown  you  how  the resulting  design  implications,  what  it  means  to  do, work  with  children  that  have  asthma, what  it  means  to  work  with  individuals  that  have  PTSD, and  how  this  ecological  approach  is  transferable. Then  that  we  can  use these  context  to  study  things  that  we wouldn't  have  thought  about  before  dime. My  student  that  graduated  last  year was  able  to  look  at  PPE, the  ecology  around  building  this  artifact  in  the  wild. Other  lessons  I've  learned  from this  intersectional  approach  is  that  for  me, the  ecological  systems  theory  really  broadens the  definition  of  what  a  caregiver  is,  right? In  autism  work  that  I  didn't  describe, we  were  able  to  use  crowd sourcing  to  build  these  interventions and  so  we  could  get  the  community  to  provide  care. Then  you  saw  that  I was  showing  you  or  sharing  with  you  some  of the  red  data  and  that  allows  us  to better  understand  how  to  care  for  veterans. In  conclusion,  I've  shown  you  that  computing  can alleviate  the  negative  impacts  of  chronic  disorders. I've  shown  you  that  health  research requires  accountability  at  the  intersection of  science  and  design. My  goal,  and  this  is  again, the  beauty  about  doing  this  empirical  work, is  that  you  can  choose  to  live  in either  the  design  space  or  in  the  science  space. But  for  me,  in  order  to  show  that  my  systems  work, I  have  to  have  that  accountability  to  science. But  again,  in  order  to  do  good  science, I  need  to  show  that  my  systems are  going  to  be  useful  and  usable. I  hope  that  one  takeaway  is  that  the  future  is  ripe with  opportunities  for  HCI  to  improve  patient  engagement, continuity  of  care,  and  point  of  care  interaction. Thank  you  very  much.  All  right, I'm  happy  to  take  any  questions. Yes,  Mike. I'm  just  trying  to  find  them  here. Yes,  this  was  work  that  was  carried  out and  this  was  all  the  grad  student  leg  work,  TJ. Basically,  they  started  taking  that  design  approach. They  started  going  to informational  sessions  about  asthma. And  in  these  sessions  they  met  nurses. With  these  nurses  were  able  to find  a  clinic  that  was actually  interested  in  helping  them understand  how  to  improve  their  system  of  workflow. Again,  give  and  take  that  you  need  to  build  a  community. At  that  point,  we  were  reading different  kinds  of  theories  and  we  really thought  about  this  idea that  in  some  of  the  interviews  that  the  students  did, one  of  the  things  that  kept  coming  up  is that  children  and  also  adults  that have  asthma  or  COPD won't  understand  that  they  have  symptoms,  right? The  doctors  will  say,  people  will  be  talking  like this  because  they  can't  breathe  and  they'll  be  like, I  have  no  problem,  right? This  idea  that  symptom  awareness  is  really critical  really  drove  this  desire  to  say, what  if  we  were  able to  raise  their  awareness  about  their  symptoms. Right?  And  again,  we  had the  pulmonologist  came  up  with,  you  know, with  the  symptom  questions  as  you saw  back  to  the  way  that  you  write  and  you  do  this  work. There  was  legal  ease, so  we  say  in  the  past  four  weeks,  right? Because  we  didn't  want  to  be accountable  if  somebody  is  going  to have  an  asthma  attack,  right? So  GT  Legal  had  to  make  sure  that the  questions or  these  prompts  were  going  to  be  appropriate. And  so  it  was  really  just  this  idea  that  we  wanted to  look  at  these  two  conditions,  right? My  personal  experience,  I know  that  I  could  probably  only  have  one  cookie, but  I'm  going  to  choose  to  have  more  knowledge isn't  enough  to  moderate  our  behavior. But  the  question  was,  if  we  did  that  combination  right, initially,  when  you  do  basic  science, you  want  to  throw  in  the  kitchen  sink. And  so  we  had  a  symptom  only  condition,  right? And  then  we  had  a  knowledge  condition to  see  what  happened. And  so  it  was  really  working with  this  model  and  trying  to understand  how  do  we  derive  those  symptom  questions. Yes,  yes, yeah,  yeah,  that's  a  great  question. Which  is,  you  know,  was  it  the  symptom? And  the  interesting  answer  was  that about  half  of  the  kids  said  yes, oh  my  goodness,  I  am  sick, I  better  take  my  med. And  there  was  about  another  40  or 30%  that  said  I'm  going  to  get  my  text  messages  anyway, I'm  going  to  use  it  as  a  reminder  to  take  my  med,  right? So  the  idea  that  through the  two  different  measures,  right, one  was  driven  by  symptom, the  other  one  was  driven  by  just  utility  of  the  SMS. We  did  have  this  improvement. Again,  the  important  thing,  we  always  say that  the  statistics  gives  us  the  what, right,  and  the  interviews  gives  us  the  why. It  was  this  interesting  thing. But  again,  I  think  the  relevant  part is  because  we  had  done  a  randomized  control  trial. Because  kids  were  randomly assigned  to  one  thing  or  the  other,  right? That  we  did  have  the  soundness  of  the  rationale and  then  that  we  were  able  to  show  that  we  could  get improved  outcomes  in  children  that  had  diabetes. And  that  was  also  a  random  control  trial? Yes, there  was  a  control  group that  got  nothing  right. So  we  had  a  we  had  a  group  that  was  assigned  to symptom  on  another  group  that was  assigned  to  symptom  and  knowledge, and  then  a  control  group  and  then  we also  had  a  group  that  was  the  knowledge  only. Right.  Yes, yeah, yeah,  yeah. So  again,  I,  you  know, I'm  drawn  to  cognitive  models because  that  was  my  training. But  you  can  imagine  that  other  people  might  be,  you  know, people  in  Europe  are  still  very,  very  inclined. It  could  be  that  people  could  use  other  types  of  models. I  use  cognitive  behavior  models because  those  are  the  ones  that  appeal most  to  me.  That's  how  I  grew  up. Intellectually.  Yes. Yes. At  feel, right.  Yeah. Right.  Again.  And  this  is  where, you  know,  if  I  had  been  in a  behavioral  science  department,  then  I  would  have. And  I  wanted  to  know, cross  all  those  IRB  hurdles  to  figure  out how  many  people  are  getting  their  prescription  refilled, right?  How  many  people  are. So  to  your  point,  definitely we  could  have  looked  at  that. But  there  were  constraints  of,  you  know, not  being  in  that  space  and  we  do  our  best,  right? All  right. So  yes, yeah,  So  I  really, I  mean,  one  of  my  personal  philosophy  is  that unless  I've  already  given something  to  the  community,  I  can't  ask  for  it. So  I  told  you  that  one  of  the  reasons that  we  were  appealing  to the  clinical  group  is  because they  wanted  to  understand  their  workflow,  right? And  so  we  did  this  study, we  were  able  to  give  them  something  then  they felt  comfortable  enough  that they  were  going  to  get  something  from  us. There  was  that  the  author, I  think,  of  our  asthma  publications and  all  of  the  other  publications have  stakeholders  in  them,  right? The  other  thing  is  we  start  early, right  now  we're  trying  to  do  work  with  this diabetic  in  a  community  housing. Well,  this  semester  and  next semester  we  are  doing  tech  help, where  we  go  out  on  Thursday  and  we're saying  if  anybody  needs any  tech  help,  we're  here  to  help  you. Eventually,  we're  going  to  ask  if  anybody  wants  to  be  in, in  our  diabetic  foot  studies, but  that's  not  going  to  be  everybody. So  how  can  we  give  something  to  the  community? We're  building  in  that  sense, we're  building  these  relationships. Mhc,  right? The  MS,  HCI  students  are  going  out to  basically  to  help  when  they  can, but  also  to  get  an  understanding  of  the  landscape. Yeah,  yes,  yes, yeah, yeah,  yeah. When  we  were  trying  to merge  or  really  to  understand  how  could  we flatten  this  idea  of the  macrosystem  and  the  exosystem and  the  mesosystem,  right? This  idea  that  we  have  added to  the  ideologies  of  a  culture, this  question  of  how  can  you  get  at  that, that's  been  a  question  that  drove  a  lot of  the  work  that  I've  done  in  autism, understanding  what  are  different  communities  values and  thoughts  about  technology  for  autism. And  so,  you  know,  you  can  say, well  you  can  go  to  meet,  for  example,  veterans. But  in  fact,  it's  incredibly  hard back  to  the  point  that  you  made  earlier. It's  incredibly  hard  to  meet  some  of  these  communities. And  so  really  read  it  gave  us  this  great  opportunity, I'm  going  to  find  it  somewhere, but  read  it  gave  us  this  great  opportunity  to  understand, you  know,  what  veterans  actually  wanted,  right? Because  back  to  thinking about  these  methods,  we  always  think, well,  it's  their  self  selection, there's  self  selection,  there's the  Hawthorne  effect,  et  cetera,  et  cetera. But  in  some  sense,  when  we  go  to  read  it, sure  it's  self  selection because  it's  the  people  that  go  there. But  on  the  other  hand,  it's  giving  us an  understanding  of  what  these  people  care  about. If  we're  trying  to  represent what  they  want  and  what  they  need. The  good  way  of  doing that  intrusively  and  when  it  comes  up making  sure  that  the  text  that  the actual  read  it  posts  don't  come  out  so  that we  can  keep  the  anonymity. But  hopefully,  again,  at  some  point, be  able  to  build  something  that's actually  useful  to  them. All  right. Yes, yeah, yeah,  yeah. I  hope  one  idea  is that  I've  always  thought  that  if  we  have  more  data, then  of  course  always  qualified,  right? That  more  data  makes  better  systems. But  we  have  to  have  the  representation  because especially  just  thinking  about  this. Now  when  we  think  about  the  images  or  we  think about  doing  these  computer  vision  stuff, there's  no  data,  there's  not enough  data  on  people  that  don't  have  just  white  skin. I  think  that  to  me  it  just  reiterates this  idea  that  we  need  to  make  sure  that we're  getting  representative  datasets. For  me,  I  think  I'm  just  constantly reminded  how  do  we  get  representative  data  sets? Back  to  the  question  of  how  do  we  start  to triangulate  on  what  veterans  want. Of  course,  it  applies  to  machine  learning, but  it  also  generally  applies  to, I  think  good  data  science  in  general.  Yeah. Well,  I  don't  want  to  keep  you. You've  been  very  generous  with  your  time. Anything  else?  All  right.  Thank  you.