Title: Scuba Diving
1Scuba Diving Example Application
2Scuba
- Scuba Self Contained Under-water Breathing
Apparatus - Scuba diving popular form of recreational
diving - 1 million divers get certified every year
- Safety of all these divers is a serious issue
- By diving carelessly, divers suffer from
decompression illness (DCI), more popularly the
bends
SCUBA Diver
3Decompression Illness (DCI)
- Possible explanation of DCI bubble theory
- Nitrogen from air absorbed by body tissue because
of the water pressure - When diver rapidly ascends the dissolved Nitrogen
forms bubbles - Large bubbles that clog joints cause pain and
bends - Individual variation of DCI
- Some don't get bent when they "should," others do
get bent when they "shouldn't."
4Dive Tables
- Traditionally dive tables are used for planning
dives to avoid risk of DCI - Dive tables contain safety limits of depth and
time - There are many standard dive tables,
- Original UK, US Navy etc.
- Dive tables can be generated by an algorithm
many variations possible - Dive computers run these algorithms and generate
dive tables on the fly
5Dive Tables
6Dive Computer (DC)
- Scuba divers wear dive computers
- DC guides the divers to carry out safe dives
- DC records logs of dives consisting of
- Log of all dives and
- For each dive the following data
- Time series of dive depth called dive profile
- Time series of Tissue saturation
- Temperature
- SCUBA equipment
- etc
- Dive logs from a DC can be downloaded to a PC
7Dive Computer Data
- DCs are equipped with software to view dive log
data as shown here - Divers are expected to analyse their dives to
learn about their safety. - For example, the dive shown here is an unsafe
dive because the diver performed a deeper dive
the second time.
8Safety of a Dive
- Is complicated to determine
- Depends at least on
- Diver characteristics such as age, general
health, history of dive related illnesses - Dive characteristics dive profile (depth-time)
data, gas mix and dive plan - Dive environment temperature and altitude
9Project Dive Exploration (PDE)
- PDE is a large scale research project sponsored
by Divers Alert Network (DAN) - PDE collects data (medical history, dive profile
etc) corresponding to large numbers of real dives
and their medical outcomes. - PDE analyses these data to learn the relationship
between features of dive data and DCI (or any
other medical condition) - PDE hopes to develop the science required to
label dives as SAFE or UNSAFE (binary classifier)
10Understanding Dive Computer Data
- PDE is ongoing and results are expected in the
future - Until then divers have to manually inspect dive
data to determine the safety of their dives - The community of divers is very diverse
- Many of them may not possess the skills required
to use the vendor supplied software - We need to help divers better understand their
dives. - We use this application as one of the example
domains in this course
11Analysis of dive data
- To determine unsafe dives
- The following patterns in dive profiles are known
to cause DCI - Rapid ascent
- Sawtooth
- Unnecessary stops
- Reverse dive profile etc.
12Segmenting a profile into zones
13Rapid Ascent
- A pattern in the dive profile caused by the diver
rising rapidly to the surface - Rapid ascent is the most critical factor causing
bubbles in body tissues - Therefore has higher chance of causing DCI
- Most dive computer software detect rapid ascents
and sound alarms
14Sawtooth
- A pattern in the dive profile caused by the diver
going down and up in quick succession - This may not happen very frequently, but when it
happens it may cause the tissues to absorb excess
gas bubbles - Therefore may cause DCI
- Dive computer software does not detect them
15Reverse Profile
- This is a pattern observed at the level of a
whole dive profile. - Ideally a diver is required to initially reach
the planned maximum depth and then all the
subsequent dive should be performed at a depth
shallower than the maximum - A reverse profile is a dive profile where the
diver performs the reverse of the ideal
16ScubaText Communication of dive data
- A research project in the department
- We explore effective ways of presenting the
results of data analysis - For scuba divers
- For scuba instructors
- For health professionals attending to divers
- For general public (dive blogs)
- Using visualizations and Text
17ScubaText Prototype
Data Analysis analysing raw data for required
features/patterns Data Interpretation mapping
the data features/patterns to the actual dive
features and rating the dive based on the dive
features e.g. long bottom times receive low rating
18Example Dive- Dive Context
- Date Mon, 04/10/1993 Location Elba
- Time 1537 Site Capo dArco
- Altitude range 0m900m Interval 0005
- Weather Cloudy Air Temp not recorded
- Dive suit two pc. Wetsuit Tank Size 14.0l
- Maximum Depth 48.0m Dive Time 0047
- Min. temperature 190C Air consumption131bar
- Dive Type Decompression, single ascent, sea
water - Activities Sightseeing
- Alarms None
- Buddies YYY
- Max ascent time 10
19TextAnnotated Graphics (D)
Risky dive with some minor problems. Because your
bottom time of 12.0min exceeds no-stop limit by
4.0min this dive is risky. But you performed the
ascent well. Your buoyancy control in the bottom
zone was poor as indicated by saw tooth
patterns marked A on the depth-time profile.
20Text and Graph
- Text mainly communicates a safety message
- Risky dive, Safe dive etc.
- Uses dive features inferred from raw dive data to
explain the main message - Links data features to dive features as further
explanation - E.g. saw tooth pattern linked to poor buoyancy
control - But not all the terms referring to data features
do not have universally accepted definitions - Bottom time, bottom zone etc
- Graph provides semantic grounding for these terms
- Text and graph are linked
- References to annotations in the graph
21Summary
- Dive computers record dive data
- Not possible to label dives SAFE or UNSAFE
automatically - Divers need help of novel technology to explore
their dives - Detect unsafe patterns
- Present the results
- Graphically
- Textually