Note extraction summary: Invented ALS cohort (demo)

Candidates in each content category

Participants with at least one candidate match in each category. A candidate is text that matched a pattern; it still needs review.

Content category	Participants with a candidate
ALSFRS-R	58
Pulmonary	58
Diagnosis	58
Family history and genetics	58
ECAS and FTD	58
Treatment milestones	56

All patterns

Context cues are checked in the words just before each match: a negation such as "no" or "negative for", a family member, uncertainty such as "possible" or "rule out", or a historical marker such as "history of". A cue does not mean the match is wrong; it marks candidates to read first.

Pattern	Category	Participants	Candidates	Share of candidates with a context cue
`alsfrs_r_total` ALSFRS-R total score, range 0 to 48	ALSFRS-R	58	197	none
`alsfrs_r_bulbar` ALSFRS-R bulbar subscore, range 0 to 12	ALSFRS-R	58	197	none
`alsfrs_r_fine_motor` ALSFRS-R fine motor (upper extremity) subscore, range 0 to 12	ALSFRS-R	58	197	none
`alsfrs_r_gross_motor` ALSFRS-R gross motor (lower extremity) subscore, range 0 to 12	ALSFRS-R	58	197	none
`alsfrs_r_respiratory` ALSFRS-R respiratory subscore, range 0 to 12	ALSFRS-R	58	197	none
`ecas_total` ECAS total, maximum 136	ECAS and FTD	58	197	none
`ftd_spectrum` Frontotemporal-spectrum involvement mention	ECAS and FTD	17	65	possible family history 54%
`el_escorial` El Escorial or Awaji-Shima diagnostic certainty	Diagnosis	46	158	none
`onset_region` Site of disease onset	Diagnosis	58	197	none
`family_history_negative` Negative family history	Family history and genetics	28	96	possible family history 100%
`family_history_positive` Affected relative with ALS, MND or FTD (positive family history)	Family history and genetics	30	101	possible family history 100%
`genetic_mutation` ALS-associated gene with a reported variant or expansion	Family history and genetics	49	68	none
`fvc_percent_predicted` Forced vital capacity, percent predicted	Pulmonary	58	197	none
`peg_placement_date` Date of PEG or gastrostomy tube placement	Treatment milestones	18	53	none
`niv_initiation_date` Date non-invasive ventilation started	Treatment milestones	28	89	none
`riluzole_start_date` Date riluzole was started	Treatment milestones	48	162	none
`edaravone_start_date` Date edaravone was started	Treatment milestones	14	46	none
`tracheostomy_date` Date of tracheostomy	Treatment milestones	held back for cell size	held back for cell size	held back for cell size

Captured values

Counts are participants, not records.

Numeric values, one per participant

Pattern	Participants	25th percentile	Median	75th percentile
`alsfrs_r_bulbar`	58	7	8	9
`alsfrs_r_fine_motor`	58	6	7	8
`alsfrs_r_gross_motor`	58	6	7	8
`alsfrs_r_respiratory`	58	6.25	8	9
`alsfrs_r_total`	58	25.25	30.5	33
`ecas_total`	58	94	101	110.75
`fvc_percent_predicted`	58	64.5	70	78

Minimum and maximum are not shown because a single extreme value belongs to one participant. The number is the first number in the captured text.

Categorical values

onset_region

Captured value	Participants
limb	19
bulbar	10
upper limb	10
lower limb	10
respiratory	9

genetic_mutation

Captured value	Participants
c9orf72	49
fus	10

el_escorial

Captured value	Participants
clinically possible	12
lab-supported probable	12
clinically probable	11
clinically definite	11

family_history_positive

Captured value	Participants
als	20
frontotemporal dementia	10

family_history_negative

Captured value	Participants
no family history of als	18
no known family history of neurodegenerative disease	10

ftd_spectrum

Captured value	Participants
frontotemporal dementia	10
behavioral variant ftd	9

About these dashboards

note_extraction.py searches C-CDA section text, decoded documents (RTF, PDF, HTML, plain text) and text attached to FHIR DiagnosticReport, DocumentReference and Binary resources. Every match is written to note_extractions.csv with its content category, a context cue and review_status needs_review.

The patterns are seed patterns for ALS registries. Check each pattern against a sample of real documents from each contributing health system, and have a reviewer confirm candidates against the source before any value is used in analysis.

Settings: k = 5. Generated 2026-11-13 12:52 UTC.