Welcome,  everyone.

Today,  I  will  be  going through  custom  heat  maps

and  how  they  provided  some new  insights  into  pet  treating.

First  of  all,  I'm  Jared  Shaw

and  I  work  for  Mars.

I've  been  with  Mars  for  eight  years.

Prior  to  Mars, I  worked  in  semiconductor I ntel

and  IM  flash  technologies for  about  13  years.

My  background  is  statistics  and  education.

I've  done  a  lot  of  consulting over  the  years  as  well  as

teaching  others in  various  statistical  methods.

I'm  married  and  I  have  three  kids.

They're  all  adopted.

I  play  games  on  the  sidelines,

build  models and  then  periodically  camping.

Then  I  also  tinker around  with  construction.

On  the  bottom  right,  there  is a  room  over  my  garage  that  I  finished.

Today,  I'll  be  going  through basically  the  abstract  I  submitted.

Then  C&T  stands  for  care  and  treat.

I'll  give  some background  on  that  for  Mars.

I'll  go  through  measuring  efficacy,

research  protocols  we  have for  these  types  of  studies.

Then  I'll  get  into  the  JMP  portion.

This  is  not  a  live demonstration  of  JMP.

I'll  just  be  showing some  images  from  the  program

and  talking  through  some  approaches that  we  used  in  looking  at  this  data

and  then  ending  up  with  the  new approach  that  I  introduced.

First  of  all,

the  abstract  here  is I  have  these  custom  heat  maps

and  they  provide  a  lot of  insights  into  pet  treating.

Overall,  our  intention  is  to  improve  the cleaning  of  pet's  teeth  with  a  new  treat.

We  do  this  by  changing  texture, changing  ingredients.

We  want  to  do  something  that  will  help impact  the  teeth,

but  also  be  safe  and  delicious

and  fun  for  the  pet as  well  as  the  pet  owner.

The  results  provide  a  lot  of  insights,

including  patterns  that  we see  across  the  mouth.

How  does  the  product  affect  the  teeth?

The  current  graphs  and  methodologies.

The  modeling  is  pretty  good,

but  the  methodologies  and  how  we  showcase this  data  is  not  very  good.

It  doesn't  offer  very  clear  insight unless  there's  a  remarkable  difference

between  the  products that  are  being  tested.

I  found  that  there  is  a  great opportunity  to  utilize  JMP  mapping

to  create  some  custom  maps.

These  new  images,

they   spawned  a  whole huge  investigation,

brought  in  some  new  associates  to  do some  great  insights  and  learning.

Just  from  a  simple  image, brought  some  great  rewards.

Maybe  to  help  bottom  or to  ground  us  to  a  baseline.

First  of  all,  let's  consider what  Care   & Treats  means.

Pet  care  consists  of  a  dry  diet and  you  also  have  wet  diets

and  then  the  Care &  Treat  components.

This  is  split  up  into  two  pieces, the  treat  and  then  the  care.

Treating  products,

they  have  a  high palatability,  they  excite  pets.

They  may  add  some  extra  nutrition

and  supplement  the  pet's  diet.

They're  used  for  training.

Positive  reward  in  training,  getting the  pet  to  respond  to  your  voice,  etc.

In  some  cases,  they're  long- lasting to  help  relieve  boredom,

reduce  destructive  behavior  and  so  forth.

The  care  products,  on  the  other  hand,

these  are  also  have  a  high  probability to  encourage  consumption.

But  these  promote  a  healthy  teeth

that  we  concentrate a  lot  on  oral  care,  reducing  bad  breath

and  they  can  also  act  as  a  medication

to  give  appeal  to  your  pet.

You  may  have  heard  of  like a  pill  pocket,  et  cetera.

Now,  one  of  the  main drivers  of  these  treats  is  texture

and  that  really  promotes the  consumption  benefits.

Does  it  become  fun  to  chew  the  product?

For  those  that  have  dogs,

you  may  notice  that  your  dog,

in  some  cases  when they  start  eating  a  product,

they  seem  to  inhale it  more  than  they  chew  it.

The  texture  piece  definitely  is  something that  you  want  to  have  them

enjoy  the  experience of  biting  into  the  product.

This  is  just  an  image to  help  us  understand

what  we're  talking  about  for  treating

and  trying  to  clean the  teeth  for  our  pets.

These  are  just  a  couple of  different  images  that  show

the  breakdown  of  the  teeth  in  pets

and  how  we  want  to  understand

how  a  product  is impacting  these  different  teeth.

How  do  we  measure  efficacy?

Efficacy  is  basically  how well  the  product  is  running.

Periodontal  disease  is  the  most widespread  of  oral  disease  in  pets.

Companies  all  over, when  they  get  into  the  care  space,

they're  looking  for  how  they  can  take texture,  how  they  can  take  shapes,

how  they  can  build  a  chew that  really  affects  this  oral  care.

They  want  to  measure the  efficacy  of  the  treats.

They  use  different approaches  over  the  years.

I'm  not  really  going to  go  too  much  into  these.

This  is  just  more  informational...

Years  ago,  there  was a  Logan  & Boyce  method.

This  is  a  visual  measure  on  the  teeth.

It  was  invasive  in  how they  did  it  against  the  pets

and  you  can  read more  about  that  on  your  own.

But  there's  also  GCPI.

This  was  less  invasive  to  the  pets, but  still  a  manual  approach.

Then  recently,  probably  within the  last  maybe  4  or 5  years,

they  came  up  with  this  QOLF  technique

that  essentially  takes  an  image of  the  teeth  before  and  after.

We  find  that  this  is just  much  more  informative.

It  gives  us  much  clearer  results on  how  things  are  proceeding  when

the  pets  consume  these  products.

Efficacy formula  for  itself.

First  of  all, there's  what's  called  an  ITS

and  you'll  see  this in  the  data  later  on.

It  stands  for  Individual  Tooth S core.

Basically,  it's  looking  at  how much  plaque  exists  on  the  tooth.

In  this  case  for  the  data  that  I'm using  is  based  upon  the  GCPI  approach.

The  length  of  the  tooth and  this  gives  us  an  idea  of  basically,

how  much  plaque  is  on each  one  of  those  teeth.

The  Chew  X,  you'll  see...

Actually,  have  them called  different  names.

But  essentially,  this  is  the  treat that's  being  tested.

Then  the  overall  efficacy,

this  is  again,  that  ability

or  can  we  produce  the intended  result  from  the  product?

This  looks  at  the  calculation takes  the  no  chew,

subtract  the  result  of  the  chew and  then  divides  by  that  no  chew

and  we  get  that  efficacy.

The  research  protocol.

The  background image  here  is  actually

one  of  our  feeding  center here  in  Tennessee.

The  round  sections,  there  are  dog  pods.

We  have  several  dogs  within each  one  of  those  pods

and  then  the  center building  has  the  cat  rooms.

What  we  do  is  we  prepare the  pets  by  cleaning  their  teeth

so  they  get  a  professional  cleaning.

We  try  to  get  all  of  plaque  off the  teeth  to  give  them  a  score  of  zero.

We  run  a  crossover  design.

Essentially,  this  means  that  every  Chew is  going  to  be  administered  to  every  pet.

Not  at  the  same  time.

We  break  it  up into  different  phases.

In  each  of  these  phases, the  chews  are  then  rotated

against  different  dogs,

as  I  have  written  here, or  cats  as  well.

Scoring  this  is  essentially done  between  each  phase.

After  a  phase  of  the  study,

so  after  those  pets, they  have  their  standard  diets.

They  get  maybe  a  treat product  at  the  end  of  the  day

and  then  at  the  end  of  whatever prescribed  time  frame,

they  measured  the  amount of  plaque  that  is  on  the  teeth.

Cleaning  teeth  with  a  score of  zero  across  all  the  treatments.

We  just  removed  these  from  the  study.

It's  just  something  that  where they're  consuming  the  product.

But  for  whatever  reason,  that  tooth didn't  get  impacted  by  the  product.

Typically,  we'll  see these  with  the  front  teeth,

the  in cisors, they're  used  more  for  cutting.

Generally,  the  products  are more  about  the  chewing  behavior.

This  whole  mouth,

what  this  is  talking  about  is  some  cases, we  have  these  individual  tooth  scores.

We  have  zero  on  a  No  Chew, so  basically  it's  missing,

or  we  find  that  the  No  Chew  has results  that  are  less  than  the  tooth.

Basically,  No  Chew  means  that

for  that  phase  of  the  study, the  dog  or  cat,

they  did  not  receive a  treat  product  to  consume,

they  just  had  a  standard  diet.

The  Chew  X  means  that  they  had some   care  treat  at  the  end  of  the  day.

We  summarize  the  data across  the  whole  mouth.

Sometimes  we'll  break  it  up  into  regions

in  order  to  give  us  an  idea of  how  it's  performing.

The  analysis  protocol  itself,  really these  are  done  with  linear  mixed  models.

We  have  fixed  effects with  maybe  treatments

or  the  regions  depending  on  the  study.

Then  we  have  random  effects that's  really  focused  around  the  pet  ID.

Again,  the  intent  here  is  that we  have  the  effects  affect  all  pets

regardless  of  the  specific pets  in  the  study.

We  also  then  run  specific  contrast.

This  is  where  we  look at  different  sizes  of  the  mouth,

different  regions,  etc .

Over  here  on  the  right- hand side  in  this  table

is  an  example  of  some of  those  contrast

and  depending  on the  number  of  contracts  we  run,

of  course,  we're  going  to  use  the  FWER,

to  control  for  inflated  error.

Then  we  communicate  these  results.

We  take  the  analysis  results,

we  take  images  and  then we  sit  down  with  the  stakeholders

and  we  show  them which  of  these  Chews

was  better  essentially.

Initially,  when  I  started  getting involved  in  these  studies,

it  was  very,  this  Chew  did  better  versus this  other  Chew  for  the  whole  mouth.

But  as  we  started  bringing in  different  regions  of  the  mouth,

we  started  seeing some  different  results

and  had  much  more fruitful  discussions.

This  will  get  us  into  the  analysis.

What  I'm  going  to  do  here  is

I'm  just  going  to  concentrate  really on  the  data  visualization  component.

I'm  not  going  to  go  too  much  into the  statistics  on  the  modeling  piece.

This  is  just  about  visualization

and  in  this  first  part is  specifically  about

ways  that  we  are  trying to  visualize  the  teeth  initially.

This  is  a  results,  this  is  from  JMP,

from  running  the  mixed  effects  model

and  then  at  the  end here  we  are  running  these  contrasts.

In  these  type  of  results as  we  look  at  these,

because  I  have  here  marked  in  the  center,

the  Chews  would  show  no  difference

but  areas  of  the  mouth  would, particularly  the  molars.

You  can  see  here  on  the  left, I  have  just  Chew  by  itself  being  compared

and  then  on  the  right- hand  side,

you  see  I  have  different areas  of  the  mouth.

Different  areas  of  mouth were  showing  interesting  differences

but  the Chew  by  themselves compared  to  No  Chew,

maybe  we're  not  seeing  too  much for  one  of  them  but  some  for  another.

Then  we  would  group  them  into  different sections  used in  the  variability  chart.

This  shows  my  different Chews  with  the  no  Chew

and  then  again the  areas  of  the  mouth.

In  this  case,  I  would  see  that the  mean  of  the  data  is  here  on  the  left

and  then  the  standard  deviation of  the  data  is  on  the  right.

Definitely,  one  area  of  the  mouth is  operating  differently

than  another,  as  I  can  see  here  on  the  right- hand  side.

Let  me  just  turn  on  my  pointer.

Over  here,  we  can  see  that  this variability  for  the  lower  molars

versus  the  upper  molars is  different  for  different  Chew.

Other  ways  that  we  tried  to  portray this  is  using  graph  builder,

we  used  the area  of  the  mouth  over  here.

I  forgot  to  mention this  earlier  but  the  IUL,

this  is  in sisors and  upper  lower  canine  teeth

and  then these  are  the  molars.

We  can  see  definitely some  pattern  going  on

when  I  compare across  different  phases.

I  have  phase  1, phase  2  and  phase  3.

Phase  3,  it  looks  like I  have  this  linear  effect  of  sorts

that's  occurring between  the  Chews.

It's  just  how  it's  showing  up  visually.

It  doesn't  mean

in  the  order  in  which  they  are  given,

it's  just  what  the  data  is  showing.

Looking  further into  the  variability  chart,

bringing  in  the  phases.

We're like  "Hey,  do  we  see differences  per  phase?"

Here  we  really see  for  this  Chew  W,

the  variability  was  much lower  than  No Chew  and  Chew P.

It  really  starts  questioning, well,  what's  going  on  here?

Why  is  this  specifically happening  for  this  Chew?

What  could  we  do to  understand  that  better?

Another  visual  that we  generated  for  this  study

is  we  again,  summarized  it by  the  area  of  the  mouth

and  then  the  Chew  efficacy for  each  of  the  Chews  themselves.

We  can  definitely  see  some differences  between  the  Chews,

but  overall,  they  might seem  like  that  they're  similar

even  though  we're  seeing differences  in  the  areas  of  the  mouth.

One  of  the  things  that I  started  asking  is  like

why  do  we  see  these  differences between  areas  of  the  mouth

but  we  don't  see  across  the  Chews as  much  what's  going  on?

Here  I  generated  a  plot where  down  here  on  the  x- axis,

I  have  the  different  dogs

and  then  the  Chew  efficacy for  the  W  and  P  Chews

and  then  areas  of  the  mouth.

Definitely, what's  interesting  here  is  that

particular  animals are  showing  the  difference

and  other  animals  are  not  now.

We  would  expect  this,

given  randomness  of  the  study  that the  Chews  are  going  to  behave  differently

and  how  the  pet is  consuming  the  product.

We  really  started  making  me  think it,

much  more  deeply  about  the  data

and  say  really  what's going  on  here  in  this  data?

Do  the  pets  chew the  product  differently?

That  led  me  into  this  data visualization  for  the  second  part

because  it  made  me  really start  thinking  about  the  data,

what  can  I  do  or  how  can I  look  at  this  differently

to  bring  out  this  individual component  of  the  pets.

I  was  working  with  a  research  scientist

and  we  were  going through  one  of  the  studies

and  they  happened  to  have  this  card.

As  you  can  see  here  on  the  right  hand side,  this  is  just  a  picture  of  the  card.

They  had  this  sitting  on  their  desk and  I  was  sitting  there  staring  at  it.

I  had  the  idea, "What  if  I  created  a  tooth  map  in  JMP?"

I  could  then  color  each  of  these individual  teeth

and  maybe  get  some  clarity,

further  clarity  in  these  studies than  what  we  were  looking  at.

I  went  and  contacted our  Waltham  scientist.

Waltham  is  a  site within  pet  care  in  the  UK

and  they  concentrate  on  doing research  on  the  pet  nutrition.

I  went  and  talked  to  them  and  one of  their  scientists  drew  me  up  some  teeth.

Here  for  this  first  slide, we  have  the  dog  teeth.

The  map  is  here  on the  left  that  they  drew  up.

On  the  right  is  just a  visual  to  give  you  an  idea

of  the  different  types  of  teeth that  show  up  in  the  dog's  mouth.

Then  we  see  something similar  for  cat  teeth.

Again,  on  the  left  is  the  one that  was  drawn  up  for  this  study.

What  I  did  then  is  this is  the  map  creator.

This  is  a  script  that's  available on  the  JMP  community.

This  is  an  older  script.

It's  been  a  while since  it's  been  updated,

but  I  found  that  it  was very  helpful  for  this  scenario

and  so  downloaded  the  add  in

and  it  creates the  add- in  pull- down  menu.

You  go  to  the  add- in  pull- down as  you  see  up  here  on  the  upper  left.

You  can  click  on  Map  Shapes  and  then you  can  do  the  Custom  Map  Creator.

When  this  pops  up,  you  get this  screen  here,  again  without  the  teeth.

Then  you  get  a  couple  of  empty  tables.

I  dragged  the  image  file  onto  the  map.

I  gave  it  a  name.

Then  I  go  over  here  to  this  next  section, and  basically,  I  start  tracing  the  teeth.

For  every  single  tooth,  I  would  trace  it and  then  I  would  give  it  a  name.

This  is  an  example  of  after doing  all  of  that  work.

As  you  can  see,  these  are  all of  the  individual plot  points,

is  me  just  clicking  around  that tooth  to  try  to  get  the  entire  shape

accurately,  so  it  would  show up  as  accurate  as  possible

on  the  screen when  we  look  at  the  plots.

Then  we  have  our  different  files  here.

You  have  this  XY.

This  gives  you  the  coordinates.

Down  here  on  the  graph  on  the  lower left,  you  see  this  is  a  zero,  zero.

This  is  essentially  like an  X,  Y  coordinate  system.

It's  just  telling  me  where  on the  graph  that  particular  point  is

for  that  particular  shape  ID.

Then  I  have  a  name  file that  gives  me  the  shape  ID

and  then  the  name  of  the  tooth.

In  this  case, I  created  a  separate  file  for  dog  teeth

and  of  course,  for  cat teeth,

since  they  are  different  shapes  of  teeth and  different  shapes  of  the  jaw.

D  just  stands  for  dog  and  then the  ID  number  for  that  tooth.

One  thing  that  I  found  interesting  is  when

I  first  created  this  program, this  was  a  few  years  ago,

I  was  able  to  just  create  the  maps, and  I  created  a  custom  script.

People  would  run  the  script,

it  would  save  the  maps  to  their  C drive  and  everything  would  work  fine.

But  soon  after  a  couple  of  years, it  no  longer  worked.

It  was  because  Mars  entered in  some  security  protocols  that

basically  wouldn't  allow  us to  save  maps  to  the  C  drive.

It  basically  locked  it  up.

I  had  to  go  out  and  figure  out, well,  how  can  I  still  do  this?

I  want  to  see  the  maps, we  want  to  create  these  maps.

Then  I  found  another  community forum  that  talked  about  putting  it  out

onto  the  app  data  for  your username  roaming,  etc .

You  see  the  path  here and  so  you  put  your  maps  there

and  it  works  just as  if  I  put  them  on  the  C  drive.

Here,  once  you  have  those files  saved  on to  their  proper  location,

then  you  go  into  JMP Graph  Builder.

If  you've  never  used  it  down  here  on the  lower  left- hand  corner  of  the  screen,

it  doesn't  show   it, it  just  says  Ma.

But  that  is  the  map  feature.

Since  I  gave  these  tooth  IDs  as  map component  or  the  name  component,

then  I  take  that  tooth  ID and  drag  it  down  to  that  section.

When  you  do,  you  can  see  here  in  the background,  I  see  those  teeth  showing  up.

Not  all  of  the  teeth  show  up because  for  this  particular  study,

I  did  not  look  at  every  single  tooth.

You  can  see  the  incisors  are  missing.

You  can  right- click  on the  image  and  go  to  Map  Shapes

and  show  the  missing  shapes.

When  you  do  that,  you  get all  of  the  teeth  showing  up.

In  addition,  I  took  the  different Chews  that  were  investigated

and  I  dragged  that  up  here to  the  Group  X  up  here  at  the  top.

We  see  these  three  different  maps for  each  of  the  Chews  and  the  No Chew.

I  then  take  the  ITS and  pulled  it  over  here  to  color.

Again,  ITS  is  the  individual  tooth  score, about  how  much  plaque  is  on  the  tooth.

At  this  point,  this  has  given  me the  average  amount  of  plaque

that  is  on  all  of  these  dog's  teeth that  were  in  this  particular  study.

I  can  start  to  see  where  that  plaque is  showing  up  on  the  No Chew .

Definitely,  it's  on  these  molars especially  down  here  on  the  bottom  molars,

on  the  right- hand  side  especially.

Then  I  can  also  see for  the  different Chews ,

I  could  see  how  maybe some  of  the  canine  teeth

on  average  was  showing  that  some of  this  plaque  remained  on  the  teeth.

But  definitely,  I'm  seeing some  cleaning  of  the  molars.

Clicking  on  done and  giving  me  the  bigger  image

so  I  can  see  it  in  more  detail.

This  leads  me  into  data  discovery.

I  created  these  maps and  they're  looking  great.

People  like,   "Hey,  this  is  a  really  interesting

way  of  looking  at  the  data."

But  I  wasn't  done  there.

We  started  discovering  something when  we  looked  at  the  maps  differently.

In  this  case,  this  is  just  back to  where  we  were, that  same  map.

What  I  did  is  I  started,

I  put  a  local  data  filter on  and  by  the  dog  names.

Here  I  have  dog  names  on  the  left, turn  on  a  local  data  filter

and  I  can  now  filter  on each  one  of  these  dogs

and  look  at  them  individually.

Now,  we  don't  have  time to  go  through  all  of  them,

but  I  wanted to  show  just  a  few  of  them.

Here  for  Aura.

What  was  interesting  for  Aura  is  that

we  noticed  that  for  this  Chew P,

that  these  lower  molars on  the  right- hand  side

weren't  really  getting  cleaned  very  well.

According  to  the  score, they  weren't  getting  cleaned  at  all.

This  started  telling  me  as  I  looked at  this,  "M an,  this  particular  pet,

Aura would  only  chew the  product  on  the  left  side."

For  this  Chew  W, we  actually  saw  a  different  signal.

She  actually  chewed the  product  that  seemed  like

more  on  both  sides  of  the  mouth.

Very   interesting  results.

A gain,  these  are  just  two different  types  of  textures

that  are  being  looked at  for  this  product.

Going  down  to  another  dog  here,  Gretchen.

She  showed  something  different.

For  Chew  P,

she  did  very  well  with  that  Chew,

but  for  Chew  W,

she  preferred  to  chew  it  more  on  the  right side  of  her  mouth  than  the  left  side.

Again,  these  are  completely two  different  animals

and  they  chew  the  product differently  depending  on  the  texture

and  their  preference  to  the  texture.

Not  all  dogs,  we  are  starting to  see  here  like  the  same  texture.

They're  very  individual.

When  I  was  doing  this, I  started  asking  friends  of  mine,

"Do  you  chew  with  one  side  of  your mouth  for  particular  products?"

Sure  enough,  as  we  started  collecting that  data,  we  found,  like  for  myself,

I  like  to  chew  nuts,  but  only on  the  left  side  of  my  mouth.

Others,  when  they  chewed  nuts, it  didn't  matter  which  side,  etc .

As  we  started  talking  about  it and  taking  record  of  it,

we  started  to  see,

"Hey,  these  pets are  consuming  product  more  like

an  individual  human  does  when  they  have preferences  in  how  they  look  at  product."

And  just  a  couple more  dogs  to  look at here.

Bagel  this  one,  the  top  teeth clean  better  than  the  bottom  teeth.

Just  fascinating  results. How  is  that  possible?

Because  the  product is  when  they're  chewing  it,

they're  biting down  into  the  product.

Your  teeth,  your  top  teeth  and  your  bottom teeth  are  sinking  into  that  material.

Why  would  we  get  certain components  showing  up  here?

Basically,  what  it  means for  this  particular  dog,  for  Bagel,

it's  just  the  rear  molars  that  Bagel was  using  to  chew  into  the  product.

The  front  molars,  who  wasn't  using  at  all.

Then  Muck,   this  is  a  great  example  of  the  dog.

Either  they're  not  consuming  the  treat  at all,  or  they're  just  inhaling  the  treat.

Those  are  some  of  the  customer calls  that  we  sometimes  get  is,

"Hey,  my  dog  is  not even  chewing  this  product.

They're  just  like  taking  it and  swallowing  it  whole."

Very  interesting  results.

A s  we  started  looking into  this  data  more  and  more,

it  really  led  us  to  believe  that,   "Hey,  we  need  a  new  product.

We  need  to  create  something  that  will really  bring  in  a  whole  mouth  clean.

A  chew  experience where  the  animal  likes  to  chew,

likes  to  really  get into  the  product  and  consume  it

and  to  have  that  efficacy  result

where  the  product is  helping  to  clean  the  teeth."

In summary,

what  I've  learned  from  this  experience  is

that  historical  studies  for  this particular  experience  were  basically

giving  an  average across  the  whole  mouth

and  it  wasn't  sufficient in  really  giving  us  a  good  idea

of  what  was  happening  with  the  product.

Viewing  these  tooth maps  by  individual  pets

started  showing  some very  interesting  results

that  really  we  couldn't  even  look  at  it

by  reaching into  the  mouth across  all  of  the  pets.

We  actually  need  to  start  looking  at  it  by  individual  and  start  classifying  it  by,

"Hey,  this  particular  treat impacted  these  teeth  only,

and  this  particular  treat impacted  these  teeth  only."

Start  classifying  it  in  that  way

so  that  we  can  start  learning  a  lot  more about  the  texture  of  the  product

and  how  it  was  consumed.

Now  these  findings,

we  started  applying  this  across all  studies,  all  historical  studies.

We  pulled  this  into  a  large  analysis

that  started  really digging  into  it  to  learn  more

from  the  history of  what  we've  done

and  how  it  affects things  moving  forward.

Of  course,  this  led  into  some new  product  development.

Here's  an  example  of  what  that  is.

Unfortunately,  I  can't  show  it  to  you.

The  image  is  protected.

It  is  not  yet  released,  but  it  is something  that  we're  investigating.

Thank  you.