This study instrumented one home
to read the patterns of daily living.

These slides go from the study's aim
to what survives of it.

1. The aim, the anchor and the team's starting points
2. The instruments
3. The view of the house, live
4. The model of activity
5. What the study could establish
6. What survives

The study set out to see changes in health
through changes in daily living.

“Develop a robust picture of daily living patterns within a specific elderly person’s home, in order to identify changes in behavior, or changes in bodily condition, which may suggest an emerging health problem.”

A baseline was to be the means: “By establishing a “normal” set of behaviors and conditions, we will recognize abnormal conditions which may signify a downturn in health.”

The publication, p. 8. It calls the study one individual's case of chronic obstructive pulmonary disease, and a pilot for the study of older people at home.

The study's anchor is a place, the home:
its sensors stay, and people come and go.

“If the person leaves the home they are out of the sensor, if guests visit they are also sensed. It is a place, the home. The sensors are fixed to things that stay at home.”

TJ's paper on the three anchors: “attending to a place foregrounds changing presence and use”.

First: TJ, 2026. Second: TJ McLeish, Computational Design, p. 8.

The team began from what motion detectors
and bed sensors could not say.

The publication's review of existing practices, pp. 2–5, which gives a source for each.

The list of what to track came from
two assessments, with signifiers added.

The publication, pp. 6–8, reprinting Lawton and Brody 1969 and Katz et al. 1970. It takes these two as the current standards and “an excellent starting point for our instrumentation” (p. 2).

The team listed eight kinds of sensor,
and used those it had already built.

The publication, pp. 8, 15: “we decided to use sensors we had already built and worked out visualization software for”. A spider is the team's own box of sensors. Weight and the accelerometer went in no installed instrument.

Six instruments went in over two visits.

First saved 19 December 2012. A Kinect on its own computer; a second computer with a Grid-EYE and two spiders.

Second visit, saved 7 January 2013. Two more spiders, in the bathroom and the bedroom, which also tried wireless.

Phase two was a plan: load cells under the bed and a chair, a Kinect in the kitchen, a power monitor on the television.

Fig. 2 · Phase one, in plan

The publication, pp. 14–16. Dates: the viewer's saved files. Team decks count “Six instrument packages, each with 8 sensors”.

The saved placements land in the rooms
the publication names, once y is negated.

Fig. 4 · The publication's plan

Fig. 5 · This page, live

The viewer's loader flipped the model's y, so an instrument saved at (x, y) sits at (x, −y) on it. The publication, p. 13: the viewer with no data loaded. The site's test holds all six to that picture.

The page's plan lines up with the first frame
of the team's film.

Fig. 6 · The film's first frame

Fig. 7 · This page, live, the same camera

The film: the viewer at work, 110 s, 10 January 2013, four instruments listed. The Grid-EYE differs: in the film it sits about 20 in from its saved place. The square of heat is one reading, the film's.

The model's first form assigns each activity
to a room and a sensor.

The publication, p. 13, with the numbers from the viewer's placements. Which of the living room's and kitchen's spiders read humidity differs between its pages: see the study's page.

Each assignment reads an activity
through something a sensor can measure.

“An observation is not an activity model. A sensor may report that a door opened, a cup moved, or someone entered a room, but none of those events alone establishes what the person was doing.”

The publication, pp. 15–16. Quotation: TJ McLeish, Modeling Activity, p. 3.

Seven of the nineteen listed items
reached an installed instrument.

The other twelve had no installed instrument,
so the model is silent on them.

Six had nothing assigned: continence, shopping, housekeeping, laundry, transportation, finances. Dressing and medication had a room, weight gain a plan, and well-being further study.

“They are necessarily partial, reflecting the limits of perception, instrumentation, available knowledge, and the assumptions embedded in their construction.”

The publication, pp. 8, 12–14; the reading is this deck's. Quotation, on models: TJ McLeish, Computational Design, p. 1.

The publication reads a few things
from its graphs.

• In the bathroom, with trial data, humidity spikes as showers.
• In the bedroom, light that tells a lamp, a ceiling light and daylight apart.
• In the foyer, a person's path through the heat, and a radiator as a fixed spot.

The publication, pp. 16, 20, 21. Its graphs are of readings from 9 to 13 January 2013. This draft shows no readings.

It ends at early analysis, with no baseline
and no change in health reported.

• It asks whether a sound in the bathroom was someone using it or other noise, and whether the lights had been left on.
• Feeding, from sound and temperature, is what “we hope to be able to determine”.
• Where someone was going, and why, is what “we may assume”.

The publication, pp. 15, 20, 21. That it reports no baseline is this deck's reading of its 33 pages. A script of January 2014 goes on to search one day of the bathroom's humidity for the shape of one rise, and nothing found by it is on record here.

Keeping the instruments connected
took daily attention.

• XBee radios “could not consistently push data more than 5 feet”.
• Bluetooth reached farther, but “the devices must be checked on a daily basis through Remote Desktop”.
• The bedroom's spider went back to a wire after two days.

The publication, p. 16.

The readings have not been found;
the model, placements, viewer and film have.

The viewer read the readings from a hosted database. TJ's drives were searched for them on 31 October 2026. Dates are file dates.

The dates are file dates, and the team's decks
describe a longer project.

The project's record, from TJ's archive. Team decks, 2013, p. 40, and 2014, p. 20.

What survives is a plan of what to sense,
without the readings to test it.

• What remains is a house, six instruments, nineteen items and the reasoning that joins them.
• If the readings are found, this view can replay them: the viewer's drawing rules are known, and the open house page uses them.
• The best lead for the readings is the team's own archive.

The conclusion is this deck's, on the sources that follow.

The sources are the team's publication,
its viewer's files and its decks.

• Iota. Iota's elder care publication, 2013, 33 pp. Quoted only where it does not name the resident.
• The team's viewer, December 2012: source, house model, saved placements. Spider firmware, 21 December 2012. A film of the viewer, 10 January 2013.
• SapientNitro. IOTA Partners Overview, 2014, pp. 19–20. A team presentation, 2013, p. 40. Method sheets, 2013, p. 4.
• McLeish, T. J. Computational Design, 2026, pp. 1, 8. Modeling Activity: Computational Design and the Opportunity for AI Applications, 2026, p. 3.
• Legorburu, G., and D. McColl. Storyscaping: Stop Creating Ads, Start Creating Worlds. Wiley, 2014, ch. 8, “In Their Shoes”, pp. 143–164, which discusses Iota's work with sensing.

The sensing, sensors, hardware and software were the Iota team's, at SapientNitro. Ann Kauth worked on the project. The resident is not named here, and no photograph or routine from the home is shown.