Skip to main content
. 2023 Dec 15;248(24):2547–2559. doi: 10.1177/15353702231214253

Table 3.

Cluster characterization of the full data set.

Cluster Count Mortality Number of ICD-10 codes ICD-10 (most frequent) Length of stay Gender (% male) Features (subclustering)
−1 1917 4.3297 7.00 (4.00, 13.00) A419 56 h 49 min 83.15
0 6457 0.0155 2.00 (2.00, 2.00) J039 20 h 10 min 45.33 BP, HR, SATS, age
1 7569 1.7043 6.00 (4.00, 10.00) R103 39 h 7 min 37.30 BP, temp, age, HR, gender, ICD-10 codes
2 1839 0.7069 26.00 (25.00, 26.00) I500 126 h 48 min 48.02 BP, temp, SATS, ICD-10 codes, AVPU, gender, HR, resp, age
3 42,423 0.0942 8.00 (5.00, 12.00) J440 47 h 04 min 3.06 BP, HR, temp, age, ICD-10 codes
4 116 1.7241 24.00 (23.00, 24.00) A419 124 h 44 min 99.14
5 2148 4.3762 24.00 (23.00, 24.00) A419 65 h 33 min 60.01 BP, HR, temp, age, gender, ICD-10 codes
6 147 5.4422 9.00 (6.00, 13.00) J189 68 h 26 min 99.32
7 39,054 2.6758 8.00 (6.00, 12.00) I251 46 h 31 min 99.40 HR, temp, age, ICD-10 codes

BP: blood pressure; HR: heart rate; SATS: hemoglobin saturation with oxygen.

Value format is mean (SD). ICD-10 (Tenth Revision of International Classification of Diseases) specifies codes for diseases and diagnoses, where for each cluster, the most frequent code corresponds to the following: cluster −1 (unclusterable patients), 4, 5 = sepsis; cluster 0 = acute tonsillitis, unspecified; cluster 1 = lower abdominal pain; cluster 3 = chronic obstructive pulmonary disease with (acute) lower respiratory infection; cluster 6 = chronic ischemic heart disease.

Bold values are the highest value per feature.